diff --git a/sys/net/if_llatbl.c b/sys/net/if_llatbl.c
index b82bc9fc104c..d2fb58974d3f 100644
--- a/sys/net/if_llatbl.c
+++ b/sys/net/if_llatbl.c
@@ -1,1121 +1,1123 @@
/*-
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
*
* Copyright (c) 2004 Luigi Rizzo, Alessandro Cerri. All rights reserved.
* Copyright (c) 2004-2008 Qing Li. All rights reserved.
* Copyright (c) 2008 Kip Macy. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include "opt_ddb.h"
#include "opt_inet.h"
#include "opt_inet6.h"
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/eventhandler.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/syslog.h>
#include <sys/sysctl.h>
#include <sys/socket.h>
#include <sys/kernel.h>
#include <sys/lock.h>
#include <sys/mutex.h>
#include <sys/rwlock.h>
#ifdef DDB
#include <ddb/ddb.h>
#endif
#include <vm/uma.h>
#include <netinet/in.h>
#include <net/if_llatbl.h>
#include <net/if.h>
#include <net/if_dl.h>
#include <net/if_var.h>
#include <net/route.h>
#include <net/route/route_ctl.h>
#include <net/route/route_debug.h>
#include <net/vnet.h>
#include <netinet/if_ether.h>
#include <netinet6/in6_var.h>
#include <netinet6/nd6.h>
MALLOC_DEFINE(M_LLTABLE, "lltable", "link level address tables");
VNET_DEFINE_STATIC(SLIST_HEAD(, lltable), lltables) =
SLIST_HEAD_INITIALIZER(lltables);
#define V_lltables VNET(lltables)
static struct rwlock lltable_list_lock;
RW_SYSINIT(lltable_list_lock, &lltable_list_lock, "lltable_list_lock");
#define LLTABLE_LIST_RLOCK() rw_rlock(&lltable_list_lock)
#define LLTABLE_LIST_RUNLOCK() rw_runlock(&lltable_list_lock)
#define LLTABLE_LIST_WLOCK() rw_wlock(&lltable_list_lock)
#define LLTABLE_LIST_WUNLOCK() rw_wunlock(&lltable_list_lock)
#define LLTABLE_LIST_LOCK_ASSERT() rw_assert(&lltable_list_lock, RA_LOCKED)
static void lltable_unlink(struct lltable *llt);
static void llentries_unlink(struct lltable *llt, struct llentries *head);
/*
* Dump lle state for a specific address family.
*/
static int
lltable_dump_af(struct lltable *llt, struct sysctl_req *wr)
{
struct epoch_tracker et;
int error;
LLTABLE_LIST_LOCK_ASSERT();
if (llt->llt_ifp->if_flags & IFF_LOOPBACK)
return (0);
error = 0;
NET_EPOCH_ENTER(et);
error = lltable_foreach_lle(llt,
(llt_foreach_cb_t *)llt->llt_dump_entry, wr);
NET_EPOCH_EXIT(et);
return (error);
}
/*
* Dump arp state for a specific address family.
*/
int
lltable_sysctl_dumparp(int af, struct sysctl_req *wr)
{
struct lltable *llt;
int error = 0;
LLTABLE_LIST_RLOCK();
SLIST_FOREACH(llt, &V_lltables, llt_link) {
if (llt->llt_af == af) {
error = lltable_dump_af(llt, wr);
if (error != 0)
goto done;
}
}
done:
LLTABLE_LIST_RUNLOCK();
return (error);
}
/*
* Common function helpers for chained hash table.
*/
/*
* Runs specified callback for each entry in @llt.
* Caller does the locking.
*
*/
static int
htable_foreach_lle(struct lltable *llt, llt_foreach_cb_t *f, void *farg)
{
struct llentry *lle, *next;
int i, error;
error = 0;
for (i = 0; i < llt->llt_hsize; i++) {
CK_LIST_FOREACH_SAFE(lle, &llt->lle_head[i], lle_next, next) {
error = f(llt, lle, farg);
if (error != 0)
break;
}
}
return (error);
}
/*
* The htable_[un]link_entry() functions return:
* 0 if the entry was (un)linked already and nothing changed,
* 1 if the entry was added/removed to/from the table, and
* -1 on error (e.g., not being able to add the entry due to limits reached).
* While the "unlink" operation should never error, callers of
* lltable_link_entry() need to check for errors and handle them.
*/
static int
htable_link_entry(struct lltable *llt, struct llentry *lle)
{
struct llentries *lleh;
uint32_t hashidx;
if ((lle->la_flags & LLE_LINKED) != 0)
return (0);
IF_AFDATA_WLOCK_ASSERT(llt->llt_ifp);
if (llt->llt_maxentries > 0 &&
llt->llt_entries >= llt->llt_maxentries)
return (-1);
hashidx = llt->llt_hash(lle, llt->llt_hsize);
lleh = &llt->lle_head[hashidx];
lle->lle_tbl = llt;
lle->lle_head = lleh;
lle->la_flags |= LLE_LINKED;
CK_LIST_INSERT_HEAD(lleh, lle, lle_next);
llt->llt_entries++;
return (1);
}
static int
htable_unlink_entry(struct llentry *lle)
{
struct lltable *llt;
if ((lle->la_flags & LLE_LINKED) == 0)
return (0);
llt = lle->lle_tbl;
IF_AFDATA_WLOCK_ASSERT(llt->llt_ifp);
KASSERT(llt->llt_entries > 0, ("%s: lltable %p (%s) entries %d <= 0",
__func__, llt, if_name(llt->llt_ifp), llt->llt_entries));
CK_LIST_REMOVE(lle, lle_next);
lle->la_flags &= ~(LLE_VALID | LLE_LINKED);
#if 0
lle->lle_tbl = NULL;
lle->lle_head = NULL;
#endif
llt->llt_entries--;
return (1);
}
struct prefix_match_data {
const struct sockaddr *addr;
const struct sockaddr *mask;
struct llentries dchain;
u_int flags;
};
static int
htable_prefix_free_cb(struct lltable *llt, struct llentry *lle, void *farg)
{
struct prefix_match_data *pmd;
pmd = (struct prefix_match_data *)farg;
if (llt->llt_match_prefix(pmd->addr, pmd->mask, pmd->flags, lle)) {
LLE_WLOCK(lle);
CK_LIST_INSERT_HEAD(&pmd->dchain, lle, lle_chain);
}
return (0);
}
static void
htable_prefix_free(struct lltable *llt, const struct sockaddr *addr,
const struct sockaddr *mask, u_int flags)
{
struct llentry *lle, *next;
struct prefix_match_data pmd;
bzero(&pmd, sizeof(pmd));
pmd.addr = addr;
pmd.mask = mask;
pmd.flags = flags;
CK_LIST_INIT(&pmd.dchain);
IF_AFDATA_WLOCK(llt->llt_ifp);
/* Push matching lles to chain */
lltable_foreach_lle(llt, htable_prefix_free_cb, &pmd);
llentries_unlink(llt, &pmd.dchain);
IF_AFDATA_WUNLOCK(llt->llt_ifp);
CK_LIST_FOREACH_SAFE(lle, &pmd.dchain, lle_chain, next)
lltable_free_entry(llt, lle);
}
static void
htable_free_tbl(struct lltable *llt)
{
free(llt->lle_head, M_LLTABLE);
free(llt, M_LLTABLE);
}
static void
llentries_unlink(struct lltable *llt, struct llentries *head)
{
struct llentry *lle, *next;
CK_LIST_FOREACH_SAFE(lle, head, lle_chain, next)
llt->llt_unlink_entry(lle);
}
/*
* Helper function used to drop all mbufs in hold queue.
*
* Returns the number of held packets, if any, that were dropped.
*/
size_t
lltable_drop_entry_queue(struct llentry *lle)
{
size_t pkts_dropped;
struct mbuf *next;
LLE_WLOCK_ASSERT(lle);
pkts_dropped = 0;
while ((lle->la_numheld > 0) && (lle->la_hold != NULL)) {
next = lle->la_hold->m_nextpkt;
m_freem(lle->la_hold);
lle->la_hold = next;
lle->la_numheld--;
pkts_dropped++;
}
KASSERT(lle->la_numheld == 0,
("%s: la_numheld %d > 0, pkts_droped %zd", __func__,
lle->la_numheld, pkts_dropped));
return (pkts_dropped);
}
void
lltable_set_entry_addr(struct ifnet *ifp, struct llentry *lle,
const char *linkhdr, size_t linkhdrsize, int lladdr_off)
{
memcpy(lle->r_linkdata, linkhdr, linkhdrsize);
lle->r_hdrlen = linkhdrsize;
lle->ll_addr = &lle->r_linkdata[lladdr_off];
lle->la_flags |= LLE_VALID;
lle->r_flags |= RLLE_VALID;
}
/*
* Acquires lltable write lock.
*
* Returns true on success, with both lltable and lle lock held.
* On failure, false is returned and lle wlock is still held.
*/
bool
lltable_acquire_wlock(struct ifnet *ifp, struct llentry *lle)
{
NET_EPOCH_ASSERT();
/* Perform real LLE update */
/* use afdata WLOCK to update fields */
LLE_WUNLOCK(lle);
IF_AFDATA_WLOCK(ifp);
LLE_WLOCK(lle);
/*
* Since we droppped LLE lock, other thread might have deleted
* this lle. Check and return
*/
if ((lle->la_flags & LLE_DELETED) != 0) {
IF_AFDATA_WUNLOCK(ifp);
return (false);
}
return (true);
}
/*
* Tries to update @lle link-level address.
* Since update requires AFDATA WLOCK, function
* drops @lle lock, acquires AFDATA lock and then acquires
* @lle lock to maintain lock order.
*
* Returns 1 on success.
*/
int
lltable_try_set_entry_addr(struct ifnet *ifp, struct llentry *lle,
const char *linkhdr, size_t linkhdrsize, int lladdr_off)
{
if (!lltable_acquire_wlock(ifp, lle))
return (0);
/* Update data */
lltable_set_entry_addr(ifp, lle, linkhdr, linkhdrsize, lladdr_off);
IF_AFDATA_WUNLOCK(ifp);
return (1);
}
/*
* Helper function used to pre-compute full/partial link-layer
* header data suitable for feeding into if_output().
*/
int
lltable_calc_llheader(struct ifnet *ifp, int family, char *lladdr,
char *buf, size_t *bufsize, int *lladdr_off)
{
struct if_encap_req ereq;
int error;
bzero(buf, *bufsize);
bzero(&ereq, sizeof(ereq));
ereq.buf = buf;
ereq.bufsize = *bufsize;
ereq.rtype = IFENCAP_LL;
ereq.family = family;
ereq.lladdr = lladdr;
ereq.lladdr_len = ifp->if_addrlen;
error = ifp->if_requestencap(ifp, &ereq);
if (error == 0) {
*bufsize = ereq.bufsize;
*lladdr_off = ereq.lladdr_off;
}
return (error);
}
/*
* Searches for the child entry matching @family inside @lle.
* Returns the entry or NULL.
*/
struct llentry *
llentry_lookup_family(struct llentry *lle, int family)
{
struct llentry *child_lle;
if (lle == NULL)
return (NULL);
CK_SLIST_FOREACH(child_lle, &lle->lle_children, lle_child_next) {
if (child_lle->r_family == family)
return (child_lle);
}
return (NULL);
}
/*
* Retrieves upper protocol family for the llentry.
* By default, all "normal" (e.g. upper_family == transport_family)
* llentries have r_family set to 0.
* Thus, use @default_family in that regard, otherwise use r_family.
*
* Returns upper protocol family
*/
int
llentry_get_upper_family(const struct llentry *lle, int default_family)
{
return (lle->r_family == 0 ? default_family : lle->r_family);
}
/*
* Prints llentry @lle data into provided buffer.
* Example: lle/inet/valid/em0/1.2.3.4
*
* Returns @buf.
*/
char *
llentry_print_buf(const struct llentry *lle, struct ifnet *ifp, int family,
char *buf, size_t bufsize)
{
#if defined(INET) || defined(INET6)
char abuf[INET6_ADDRSTRLEN];
#endif
const char *valid = (lle->r_flags & RLLE_VALID) ? "valid" : "no_l2";
const char *upper_str = rib_print_family(llentry_get_upper_family(lle, family));
switch (family) {
#ifdef INET
case AF_INET:
inet_ntop(AF_INET, &lle->r_l3addr.addr4, abuf, sizeof(abuf));
snprintf(buf, bufsize, "lle/%s/%s/%s/%s", upper_str,
valid, if_name(ifp), abuf);
break;
#endif
#ifdef INET6
case AF_INET6:
inet_ntop(AF_INET6, &lle->r_l3addr.addr6, abuf, sizeof(abuf));
snprintf(buf, bufsize, "lle/%s/%s/%s/%s", upper_str,
valid, if_name(ifp), abuf);
break;
#endif
default:
snprintf(buf, bufsize, "lle/%s/%s/%s/????", upper_str,
valid, if_name(ifp));
break;
}
return (buf);
}
char *
llentry_print_buf_lltable(const struct llentry *lle, char *buf, size_t bufsize)
{
struct lltable *tbl = lle->lle_tbl;
return (llentry_print_buf(lle, lltable_get_ifp(tbl), lltable_get_af(tbl), buf, bufsize));
}
/*
* Requests feedback from the datapath.
* First packet using @lle should result in
* setting r_skip_req back to 0 and updating
* lle_hittime to the current time_uptime.
*/
void
llentry_request_feedback(struct llentry *lle)
{
struct llentry *child_lle;
LLE_REQ_LOCK(lle);
lle->r_skip_req = 1;
LLE_REQ_UNLOCK(lle);
CK_SLIST_FOREACH(child_lle, &lle->lle_children, lle_child_next) {
LLE_REQ_LOCK(child_lle);
child_lle->r_skip_req = 1;
LLE_REQ_UNLOCK(child_lle);
}
}
/*
* Updates the lle state to mark it has been used
* and record the time.
* Used by the llentry_provide_feedback() wrapper.
*/
void
llentry_mark_used(struct llentry *lle)
{
LLE_REQ_LOCK(lle);
lle->r_skip_req = 0;
lle->lle_hittime = time_uptime;
LLE_REQ_UNLOCK(lle);
}
/*
* Fetches the time when lle was used.
* Return 0 if the entry was not used, relevant time_uptime
* otherwise.
*/
static time_t
llentry_get_hittime_raw(struct llentry *lle)
{
time_t lle_hittime = 0;
LLE_REQ_LOCK(lle);
if ((lle->r_skip_req == 0) && (lle_hittime < lle->lle_hittime))
lle_hittime = lle->lle_hittime;
LLE_REQ_UNLOCK(lle);
return (lle_hittime);
}
time_t
llentry_get_hittime(struct llentry *lle)
{
time_t lle_hittime = 0;
struct llentry *child_lle;
lle_hittime = llentry_get_hittime_raw(lle);
CK_SLIST_FOREACH(child_lle, &lle->lle_children, lle_child_next) {
time_t hittime = llentry_get_hittime_raw(child_lle);
if (hittime > lle_hittime)
lle_hittime = hittime;
}
return (lle_hittime);
}
/*
* Update link-layer header for given @lle after
* interface lladdr was changed.
*/
static int
llentry_update_ifaddr(struct lltable *llt, struct llentry *lle, void *farg)
{
struct ifnet *ifp;
u_char linkhdr[LLE_MAX_LINKHDR];
size_t linkhdrsize;
u_char *lladdr;
int lladdr_off;
ifp = (struct ifnet *)farg;
lladdr = lle->ll_addr;
LLE_WLOCK(lle);
if ((lle->la_flags & LLE_VALID) == 0) {
LLE_WUNLOCK(lle);
return (0);
}
if ((lle->la_flags & LLE_IFADDR) != 0)
lladdr = IF_LLADDR(ifp);
linkhdrsize = sizeof(linkhdr);
lltable_calc_llheader(ifp, llt->llt_af, lladdr, linkhdr, &linkhdrsize,
&lladdr_off);
memcpy(lle->r_linkdata, linkhdr, linkhdrsize);
LLE_WUNLOCK(lle);
return (0);
}
/*
* Update all calculated headers for given @llt
*/
void
lltable_update_ifaddr(struct lltable *llt)
{
if (llt->llt_ifp->if_flags & IFF_LOOPBACK)
return;
IF_AFDATA_WLOCK(llt->llt_ifp);
lltable_foreach_lle(llt, llentry_update_ifaddr, llt->llt_ifp);
IF_AFDATA_WUNLOCK(llt->llt_ifp);
}
/*
*
* Performs generic cleanup routines and frees lle.
*
* Called for non-linked entries, with callouts and
* other AF-specific cleanups performed.
*
* @lle must be passed WLOCK'ed
*
* Returns the number of held packets, if any, that were dropped.
*/
size_t
llentry_free(struct llentry *lle)
{
size_t pkts_dropped;
LLE_WLOCK_ASSERT(lle);
KASSERT((lle->la_flags & LLE_LINKED) == 0, ("freeing linked lle"));
pkts_dropped = lltable_drop_entry_queue(lle);
/* cancel timer */
if (callout_stop(&lle->lle_timer) > 0)
LLE_REMREF(lle);
LLE_FREE_LOCKED(lle);
return (pkts_dropped);
}
/*
* Free all entries from given table and free itself.
*/
static int
lltable_free_cb(struct lltable *llt, struct llentry *lle, void *farg)
{
struct llentries *dchain;
dchain = (struct llentries *)farg;
LLE_WLOCK(lle);
CK_LIST_INSERT_HEAD(dchain, lle, lle_chain);
return (0);
}
/*
* Free all entries from given table and free itself.
*/
void
lltable_free(struct lltable *llt)
{
struct llentry *lle, *next;
struct llentries dchain;
KASSERT(llt != NULL, ("%s: llt is NULL", __func__));
lltable_unlink(llt);
CK_LIST_INIT(&dchain);
IF_AFDATA_WLOCK(llt->llt_ifp);
/* Push all lles to @dchain */
lltable_foreach_lle(llt, lltable_free_cb, &dchain);
llentries_unlink(llt, &dchain);
IF_AFDATA_WUNLOCK(llt->llt_ifp);
CK_LIST_FOREACH_SAFE(lle, &dchain, lle_chain, next) {
llentry_free(lle);
}
KASSERT(llt->llt_entries == 0, ("%s: lltable %p (%s) entires not 0: %d",
__func__, llt, llt->llt_ifp->if_xname, llt->llt_entries));
llt->llt_free_tbl(llt);
}
/*
* Deletes an address from given lltable.
* Used for userland interaction to remove
* individual entries. Skips entries added by OS.
*/
int
lltable_delete_addr(struct lltable *llt, u_int flags,
const struct sockaddr *l3addr)
{
struct llentry *lle;
struct ifnet *ifp;
ifp = llt->llt_ifp;
IF_AFDATA_WLOCK(ifp);
lle = lla_lookup(llt, LLE_SF(l3addr->sa_family, LLE_EXCLUSIVE), l3addr);
if (lle == NULL) {
IF_AFDATA_WUNLOCK(ifp);
return (ENOENT);
}
if ((lle->la_flags & LLE_IFADDR) != 0 && (flags & LLE_IFADDR) == 0) {
IF_AFDATA_WUNLOCK(ifp);
LLE_WUNLOCK(lle);
return (EPERM);
}
lltable_unlink_entry(llt, lle);
IF_AFDATA_WUNLOCK(ifp);
llt->llt_delete_entry(llt, lle);
return (0);
}
void
lltable_prefix_free(int af, struct sockaddr *addr, struct sockaddr *mask,
u_int flags)
{
struct lltable *llt;
LLTABLE_LIST_RLOCK();
SLIST_FOREACH(llt, &V_lltables, llt_link) {
if (llt->llt_af != af)
continue;
llt->llt_prefix_free(llt, addr, mask, flags);
}
LLTABLE_LIST_RUNLOCK();
}
struct lltable *
lltable_allocate_htbl(uint32_t hsize)
{
struct lltable *llt;
int i;
llt = malloc(sizeof(struct lltable), M_LLTABLE, M_WAITOK | M_ZERO);
llt->llt_hsize = hsize;
llt->lle_head = malloc(sizeof(struct llentries) * hsize,
M_LLTABLE, M_WAITOK | M_ZERO);
for (i = 0; i < llt->llt_hsize; i++)
CK_LIST_INIT(&llt->lle_head[i]);
/* Set some default callbacks */
llt->llt_link_entry = htable_link_entry;
llt->llt_unlink_entry = htable_unlink_entry;
llt->llt_prefix_free = htable_prefix_free;
llt->llt_foreach_entry = htable_foreach_lle;
llt->llt_free_tbl = htable_free_tbl;
return (llt);
}
/*
* Links lltable to global llt list.
*/
void
lltable_link(struct lltable *llt)
{
LLTABLE_LIST_WLOCK();
SLIST_INSERT_HEAD(&V_lltables, llt, llt_link);
LLTABLE_LIST_WUNLOCK();
}
static void
lltable_unlink(struct lltable *llt)
{
LLTABLE_LIST_WLOCK();
SLIST_REMOVE(&V_lltables, llt, lltable, llt_link);
LLTABLE_LIST_WUNLOCK();
}
/*
* Gets interface @ifp lltable for the specified @family
*/
struct lltable *
lltable_get(struct ifnet *ifp, int family)
{
switch (family) {
#ifdef INET
case AF_INET:
return (in_lltable_get(ifp));
#endif
#ifdef INET6
case AF_INET6:
return (in6_lltable_get(ifp));
#endif
}
return (NULL);
}
/*
* External methods used by lltable consumers
*/
int
lltable_foreach_lle(struct lltable *llt, llt_foreach_cb_t *f, void *farg)
{
return (llt->llt_foreach_entry(llt, f, farg));
}
struct llentry *
lltable_alloc_entry(struct lltable *llt, u_int flags,
const struct sockaddr *l3addr)
{
return (llt->llt_alloc_entry(llt, flags, l3addr));
}
void
lltable_free_entry(struct lltable *llt, struct llentry *lle)
{
llt->llt_free_entry(llt, lle);
}
int
lltable_link_entry(struct lltable *llt, struct llentry *lle)
{
return (llt->llt_link_entry(llt, lle));
}
void
lltable_link_child_entry(struct llentry *lle, struct llentry *child_lle)
{
child_lle->lle_parent = lle;
child_lle->lle_tbl = lle->lle_tbl;
child_lle->la_flags |= LLE_LINKED;
CK_SLIST_INSERT_HEAD(&lle->lle_children, child_lle, lle_child_next);
}
void
lltable_unlink_child_entry(struct llentry *child_lle)
{
struct llentry *lle = child_lle->lle_parent;
child_lle->la_flags &= ~LLE_LINKED;
child_lle->lle_parent = NULL;
CK_SLIST_REMOVE(&lle->lle_children, child_lle, llentry, lle_child_next);
}
int
lltable_unlink_entry(struct lltable *llt, struct llentry *lle)
{
return (llt->llt_unlink_entry(lle));
}
void
lltable_fill_sa_entry(const struct llentry *lle, struct sockaddr *sa)
{
struct lltable *llt;
llt = lle->lle_tbl;
llt->llt_fill_sa_entry(lle, sa);
}
struct ifnet *
lltable_get_ifp(const struct lltable *llt)
{
return (llt->llt_ifp);
}
int
lltable_get_af(const struct lltable *llt)
{
return (llt->llt_af);
}
/*
* Called in route_output when rtm_flags contains RTF_LLDATA.
*/
int
lla_rt_output(struct rt_msghdr *rtm, struct rt_addrinfo *info)
{
struct sockaddr_dl *dl =
(struct sockaddr_dl *)info->rti_info[RTAX_GATEWAY];
struct sockaddr *dst = (struct sockaddr *)info->rti_info[RTAX_DST];
struct ifnet *ifp;
struct lltable *llt;
struct llentry *lle, *lle_tmp;
uint8_t linkhdr[LLE_MAX_LINKHDR];
size_t linkhdrsize;
int lladdr_off;
u_int laflags = 0;
int error;
if (dl == NULL || dl->sdl_family != AF_LINK)
return (EINVAL);
/* XXX: should be ntohs() */
ifp = ifnet_byindex(dl->sdl_index);
if (ifp == NULL) {
log(LOG_INFO, "%s: invalid ifp (sdl_index %d)\n",
__func__, dl->sdl_index);
return EINVAL;
}
llt = lltable_get(ifp, dst->sa_family);
if (llt == NULL)
return (ESRCH);
error = 0;
switch (rtm->rtm_type) {
case RTM_ADD:
/* Add static LLE */
laflags = 0;
if (rtm->rtm_rmx.rmx_expire == 0)
laflags = LLE_STATIC;
lle = lltable_alloc_entry(llt, laflags, dst);
if (lle == NULL)
return (ENOMEM);
linkhdrsize = sizeof(linkhdr);
if (lltable_calc_llheader(ifp, dst->sa_family, LLADDR(dl),
- linkhdr, &linkhdrsize, &lladdr_off) != 0)
+ linkhdr, &linkhdrsize, &lladdr_off) != 0) {
+ lltable_free_entry(llt, lle);
return (EINVAL);
+ }
lltable_set_entry_addr(ifp, lle, linkhdr, linkhdrsize,
lladdr_off);
if ((rtm->rtm_flags & RTF_ANNOUNCE))
lle->la_flags |= LLE_PUB;
lle->la_expire = rtm->rtm_rmx.rmx_expire;
laflags = lle->la_flags;
/* Try to link new entry */
lle_tmp = NULL;
IF_AFDATA_WLOCK(ifp);
LLE_WLOCK(lle);
lle_tmp = lla_lookup(llt, LLE_EXCLUSIVE, dst);
if (lle_tmp != NULL) {
/* Check if we are trying to replace immutable entry */
if ((lle_tmp->la_flags & LLE_IFADDR) != 0) {
IF_AFDATA_WUNLOCK(ifp);
LLE_WUNLOCK(lle_tmp);
lltable_free_entry(llt, lle);
return (EPERM);
}
/* Unlink existing entry from table */
lltable_unlink_entry(llt, lle_tmp);
}
lltable_link_entry(llt, lle);
IF_AFDATA_WUNLOCK(ifp);
if (lle_tmp != NULL) {
EVENTHANDLER_INVOKE(lle_event, lle_tmp,LLENTRY_EXPIRED);
lltable_free_entry(llt, lle_tmp);
}
/*
* By invoking LLE handler here we might get
* two events on static LLE entry insertion
* in routing socket. However, since we might have
* other subscribers we need to generate this event.
*/
EVENTHANDLER_INVOKE(lle_event, lle, LLENTRY_RESOLVED);
LLE_WUNLOCK(lle);
#ifdef INET
/* gratuitous ARP */
if ((laflags & LLE_PUB) && dst->sa_family == AF_INET)
arprequest(ifp,
&((struct sockaddr_in *)dst)->sin_addr,
&((struct sockaddr_in *)dst)->sin_addr,
(u_char *)LLADDR(dl));
#endif
break;
case RTM_DELETE:
return (lltable_delete_addr(llt, 0, dst));
default:
error = EINVAL;
}
return (error);
}
#ifdef DDB
struct llentry_sa {
struct llentry base;
struct sockaddr l3_addr;
};
static void
llatbl_lle_show(struct llentry_sa *la)
{
struct llentry *lle;
uint8_t octet[6];
lle = &la->base;
db_printf("lle=%p\n", lle);
db_printf(" lle_next=%p\n", lle->lle_next.cle_next);
db_printf(" lle_lock=%p\n", &lle->lle_lock);
db_printf(" lle_tbl=%p\n", lle->lle_tbl);
db_printf(" lle_head=%p\n", lle->lle_head);
db_printf(" la_hold=%p\n", lle->la_hold);
db_printf(" la_numheld=%d\n", lle->la_numheld);
db_printf(" la_expire=%ju\n", (uintmax_t)lle->la_expire);
db_printf(" la_flags=0x%04x\n", lle->la_flags);
db_printf(" la_asked=%u\n", lle->la_asked);
db_printf(" la_preempt=%u\n", lle->la_preempt);
db_printf(" ln_state=%d\n", lle->ln_state);
db_printf(" ln_router=%u\n", lle->ln_router);
db_printf(" ln_ntick=%ju\n", (uintmax_t)lle->ln_ntick);
db_printf(" lle_refcnt=%d\n", lle->lle_refcnt);
bcopy(lle->ll_addr, octet, sizeof(octet));
db_printf(" ll_addr=%02x:%02x:%02x:%02x:%02x:%02x\n",
octet[0], octet[1], octet[2], octet[3], octet[4], octet[5]);
db_printf(" lle_timer=%p\n", &lle->lle_timer);
switch (la->l3_addr.sa_family) {
#ifdef INET
case AF_INET:
{
struct sockaddr_in *sin;
char l3s[INET_ADDRSTRLEN];
sin = (struct sockaddr_in *)&la->l3_addr;
inet_ntoa_r(sin->sin_addr, l3s);
db_printf(" l3_addr=%s\n", l3s);
break;
}
#endif
#ifdef INET6
case AF_INET6:
{
struct sockaddr_in6 *sin6;
char l3s[INET6_ADDRSTRLEN];
sin6 = (struct sockaddr_in6 *)&la->l3_addr;
ip6_sprintf(l3s, &sin6->sin6_addr);
db_printf(" l3_addr=%s\n", l3s);
break;
}
#endif
default:
db_printf(" l3_addr=N/A (af=%d)\n", la->l3_addr.sa_family);
break;
}
}
DB_SHOW_COMMAND(llentry, db_show_llentry)
{
if (!have_addr) {
db_printf("usage: show llentry <struct llentry *>\n");
return;
}
llatbl_lle_show((struct llentry_sa *)addr);
}
static void
llatbl_llt_show(struct lltable *llt)
{
int i;
struct llentry *lle;
db_printf("llt=%p llt_af=%d llt_ifp=%p\n",
llt, llt->llt_af, llt->llt_ifp);
for (i = 0; i < llt->llt_hsize; i++) {
CK_LIST_FOREACH(lle, &llt->lle_head[i], lle_next) {
llatbl_lle_show((struct llentry_sa *)lle);
if (db_pager_quit)
return;
}
}
}
DB_SHOW_COMMAND(lltable, db_show_lltable)
{
if (!have_addr) {
db_printf("usage: show lltable <struct lltable *>\n");
return;
}
llatbl_llt_show((struct lltable *)addr);
}
DB_SHOW_ALL_COMMAND(lltables, db_show_all_lltables)
{
VNET_ITERATOR_DECL(vnet_iter);
struct lltable *llt;
VNET_FOREACH(vnet_iter) {
CURVNET_SET_QUIET(vnet_iter);
#ifdef VIMAGE
db_printf("vnet=%p\n", curvnet);
#endif
SLIST_FOREACH(llt, &V_lltables, llt_link) {
db_printf("llt=%p llt_af=%d llt_ifp=%p(%s)\n",
llt, llt->llt_af, llt->llt_ifp,
(llt->llt_ifp != NULL) ?
llt->llt_ifp->if_xname : "?");
if (have_addr && addr != 0) /* verbose */
llatbl_llt_show(llt);
if (db_pager_quit) {
CURVNET_RESTORE();
return;
}
}
CURVNET_RESTORE();
}
}
#endif
diff --git a/sys/netinet/in.c b/sys/netinet/in.c
index 2303a8f10aa4..92b9c229ea5b 100644
--- a/sys/netinet/in.c
+++ b/sys/netinet/in.c
@@ -1,1730 +1,1730 @@
/*-
* SPDX-License-Identifier: BSD-3-Clause
*
* Copyright (c) 1982, 1986, 1991, 1993
* The Regents of the University of California. All rights reserved.
* Copyright (C) 2001 WIDE Project. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* @(#)in.c 8.4 (Berkeley) 1/9/95
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#define IN_HISTORICAL_NETS /* include class masks */
#include <sys/param.h>
#include <sys/eventhandler.h>
#include <sys/systm.h>
#include <sys/sockio.h>
#include <sys/malloc.h>
#include <sys/priv.h>
#include <sys/socket.h>
#include <sys/jail.h>
#include <sys/kernel.h>
#include <sys/lock.h>
#include <sys/proc.h>
#include <sys/rmlock.h>
#include <sys/sysctl.h>
#include <sys/syslog.h>
#include <sys/sx.h>
#include <net/if.h>
#include <net/if_var.h>
#include <net/if_arp.h>
#include <net/if_dl.h>
#include <net/if_llatbl.h>
#include <net/if_types.h>
#include <net/route.h>
#include <net/route/nhop.h>
#include <net/route/route_ctl.h>
#include <net/vnet.h>
#include <netinet/if_ether.h>
#include <netinet/in.h>
#include <netinet/in_fib.h>
#include <netinet/in_var.h>
#include <netinet/in_pcb.h>
#include <netinet/ip_var.h>
#include <netinet/ip_carp.h>
#include <netinet/igmp_var.h>
#include <netinet/udp.h>
#include <netinet/udp_var.h>
static int in_aifaddr_ioctl(u_long, caddr_t, struct ifnet *, struct thread *);
static int in_difaddr_ioctl(u_long, caddr_t, struct ifnet *, struct thread *);
static int in_gifaddr_ioctl(u_long, caddr_t, struct ifnet *, struct thread *);
static void in_socktrim(struct sockaddr_in *);
static void in_purgemaddrs(struct ifnet *);
static bool ia_need_loopback_route(const struct in_ifaddr *);
VNET_DEFINE_STATIC(int, nosameprefix);
#define V_nosameprefix VNET(nosameprefix)
SYSCTL_INT(_net_inet_ip, OID_AUTO, no_same_prefix, CTLFLAG_VNET | CTLFLAG_RW,
&VNET_NAME(nosameprefix), 0,
"Refuse to create same prefixes on different interfaces");
VNET_DEFINE_STATIC(bool, broadcast_lowest);
#define V_broadcast_lowest VNET(broadcast_lowest)
SYSCTL_BOOL(_net_inet_ip, OID_AUTO, broadcast_lowest, CTLFLAG_VNET | CTLFLAG_RW,
&VNET_NAME(broadcast_lowest), 0,
"Treat lowest address on a subnet (host 0) as broadcast");
VNET_DECLARE(struct inpcbinfo, ripcbinfo);
#define V_ripcbinfo VNET(ripcbinfo)
static struct sx in_control_sx;
SX_SYSINIT(in_control_sx, &in_control_sx, "in_control");
/*
* Return 1 if an internet address is for a ``local'' host
* (one to which we have a connection).
*/
int
in_localaddr(struct in_addr in)
{
struct rm_priotracker in_ifa_tracker;
u_long i = ntohl(in.s_addr);
struct in_ifaddr *ia;
IN_IFADDR_RLOCK(&in_ifa_tracker);
CK_STAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
if ((i & ia->ia_subnetmask) == ia->ia_subnet) {
IN_IFADDR_RUNLOCK(&in_ifa_tracker);
return (1);
}
}
IN_IFADDR_RUNLOCK(&in_ifa_tracker);
return (0);
}
/*
* Return 1 if an internet address is for the local host and configured
* on one of its interfaces.
*/
int
in_localip(struct in_addr in)
{
struct rm_priotracker in_ifa_tracker;
struct in_ifaddr *ia;
IN_IFADDR_RLOCK(&in_ifa_tracker);
LIST_FOREACH(ia, INADDR_HASH(in.s_addr), ia_hash) {
if (IA_SIN(ia)->sin_addr.s_addr == in.s_addr) {
IN_IFADDR_RUNLOCK(&in_ifa_tracker);
return (1);
}
}
IN_IFADDR_RUNLOCK(&in_ifa_tracker);
return (0);
}
/*
* Return 1 if an internet address is configured on an interface.
*/
int
in_ifhasaddr(struct ifnet *ifp, struct in_addr in)
{
struct ifaddr *ifa;
struct in_ifaddr *ia;
NET_EPOCH_ASSERT();
CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
if (ifa->ifa_addr->sa_family != AF_INET)
continue;
ia = (struct in_ifaddr *)ifa;
if (ia->ia_addr.sin_addr.s_addr == in.s_addr)
return (1);
}
return (0);
}
/*
* Return a reference to the interface address which is different to
* the supplied one but with same IP address value.
*/
static struct in_ifaddr *
in_localip_more(struct in_ifaddr *original_ia)
{
struct rm_priotracker in_ifa_tracker;
in_addr_t original_addr = IA_SIN(original_ia)->sin_addr.s_addr;
uint32_t original_fib = original_ia->ia_ifa.ifa_ifp->if_fib;
struct in_ifaddr *ia;
IN_IFADDR_RLOCK(&in_ifa_tracker);
LIST_FOREACH(ia, INADDR_HASH(original_addr), ia_hash) {
in_addr_t addr = IA_SIN(ia)->sin_addr.s_addr;
uint32_t fib = ia->ia_ifa.ifa_ifp->if_fib;
if (!V_rt_add_addr_allfibs && (original_fib != fib))
continue;
if ((original_ia != ia) && (original_addr == addr)) {
ifa_ref(&ia->ia_ifa);
IN_IFADDR_RUNLOCK(&in_ifa_tracker);
return (ia);
}
}
IN_IFADDR_RUNLOCK(&in_ifa_tracker);
return (NULL);
}
/*
* Tries to find first IPv4 address in the provided fib.
* Prefers non-loopback addresses and return loopback IFF
* @loopback_ok is set.
*
* Returns ifa or NULL.
*/
struct in_ifaddr *
in_findlocal(uint32_t fibnum, bool loopback_ok)
{
struct rm_priotracker in_ifa_tracker;
struct in_ifaddr *ia = NULL, *ia_lo = NULL;
NET_EPOCH_ASSERT();
IN_IFADDR_RLOCK(&in_ifa_tracker);
CK_STAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
uint32_t ia_fib = ia->ia_ifa.ifa_ifp->if_fib;
if (!V_rt_add_addr_allfibs && (fibnum != ia_fib))
continue;
if (!IN_LOOPBACK(ntohl(IA_SIN(ia)->sin_addr.s_addr)))
break;
if (loopback_ok)
ia_lo = ia;
}
IN_IFADDR_RUNLOCK(&in_ifa_tracker);
if (ia == NULL)
ia = ia_lo;
return (ia);
}
/*
* Determine whether an IP address is in a reserved set of addresses
* that may not be forwarded, or whether datagrams to that destination
* may be forwarded.
*/
int
in_canforward(struct in_addr in)
{
u_long i = ntohl(in.s_addr);
if (IN_EXPERIMENTAL(i) || IN_MULTICAST(i) || IN_LINKLOCAL(i) ||
IN_ZERONET(i) || IN_LOOPBACK(i))
return (0);
return (1);
}
/*
* Trim a mask in a sockaddr
*/
static void
in_socktrim(struct sockaddr_in *ap)
{
char *cplim = (char *) &ap->sin_addr;
char *cp = (char *) (&ap->sin_addr + 1);
ap->sin_len = 0;
while (--cp >= cplim)
if (*cp) {
(ap)->sin_len = cp - (char *) (ap) + 1;
break;
}
}
/*
* Generic internet control operations (ioctl's).
*/
int
in_control(struct socket *so, u_long cmd, caddr_t data, struct ifnet *ifp,
struct thread *td)
{
struct ifreq *ifr = (struct ifreq *)data;
struct sockaddr_in *addr = (struct sockaddr_in *)&ifr->ifr_addr;
struct epoch_tracker et;
struct ifaddr *ifa;
struct in_ifaddr *ia;
int error;
if (ifp == NULL)
return (EADDRNOTAVAIL);
/*
* Filter out 4 ioctls we implement directly. Forward the rest
* to specific functions and ifp->if_ioctl().
*/
switch (cmd) {
case SIOCGIFADDR:
case SIOCGIFBRDADDR:
case SIOCGIFDSTADDR:
case SIOCGIFNETMASK:
break;
case SIOCGIFALIAS:
sx_xlock(&in_control_sx);
error = in_gifaddr_ioctl(cmd, data, ifp, td);
sx_xunlock(&in_control_sx);
return (error);
case SIOCDIFADDR:
sx_xlock(&in_control_sx);
error = in_difaddr_ioctl(cmd, data, ifp, td);
sx_xunlock(&in_control_sx);
return (error);
case OSIOCAIFADDR: /* 9.x compat */
case SIOCAIFADDR:
sx_xlock(&in_control_sx);
error = in_aifaddr_ioctl(cmd, data, ifp, td);
sx_xunlock(&in_control_sx);
return (error);
case SIOCSIFADDR:
case SIOCSIFBRDADDR:
case SIOCSIFDSTADDR:
case SIOCSIFNETMASK:
/* We no longer support that old commands. */
return (EINVAL);
default:
if (ifp->if_ioctl == NULL)
return (EOPNOTSUPP);
return ((*ifp->if_ioctl)(ifp, cmd, data));
}
if (addr->sin_addr.s_addr != INADDR_ANY &&
prison_check_ip4(td->td_ucred, &addr->sin_addr) != 0)
return (EADDRNOTAVAIL);
/*
* Find address for this interface, if it exists. If an
* address was specified, find that one instead of the
* first one on the interface, if possible.
*/
NET_EPOCH_ENTER(et);
CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
if (ifa->ifa_addr->sa_family != AF_INET)
continue;
ia = (struct in_ifaddr *)ifa;
if (ia->ia_addr.sin_addr.s_addr == addr->sin_addr.s_addr)
break;
}
if (ifa == NULL)
CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link)
if (ifa->ifa_addr->sa_family == AF_INET) {
ia = (struct in_ifaddr *)ifa;
if (prison_check_ip4(td->td_ucred,
&ia->ia_addr.sin_addr) == 0)
break;
}
if (ifa == NULL) {
NET_EPOCH_EXIT(et);
return (EADDRNOTAVAIL);
}
error = 0;
switch (cmd) {
case SIOCGIFADDR:
*addr = ia->ia_addr;
break;
case SIOCGIFBRDADDR:
if ((ifp->if_flags & IFF_BROADCAST) == 0) {
error = EINVAL;
break;
}
*addr = ia->ia_broadaddr;
break;
case SIOCGIFDSTADDR:
if ((ifp->if_flags & IFF_POINTOPOINT) == 0) {
error = EINVAL;
break;
}
*addr = ia->ia_dstaddr;
break;
case SIOCGIFNETMASK:
*addr = ia->ia_sockmask;
break;
}
NET_EPOCH_EXIT(et);
return (error);
}
static int
in_aifaddr_ioctl(u_long cmd, caddr_t data, struct ifnet *ifp, struct thread *td)
{
const struct in_aliasreq *ifra = (struct in_aliasreq *)data;
const struct sockaddr_in *addr = &ifra->ifra_addr;
const struct sockaddr_in *broadaddr = &ifra->ifra_broadaddr;
const struct sockaddr_in *mask = &ifra->ifra_mask;
const struct sockaddr_in *dstaddr = &ifra->ifra_dstaddr;
const int vhid = (cmd == SIOCAIFADDR) ? ifra->ifra_vhid : 0;
struct epoch_tracker et;
struct ifaddr *ifa;
struct in_ifaddr *ia;
bool iaIsFirst;
int error = 0;
error = priv_check(td, PRIV_NET_ADDIFADDR);
if (error)
return (error);
/*
* ifra_addr must be present and be of INET family.
* ifra_broadaddr/ifra_dstaddr and ifra_mask are optional.
*/
if (addr->sin_len != sizeof(struct sockaddr_in) ||
addr->sin_family != AF_INET)
return (EINVAL);
if (broadaddr->sin_len != 0 &&
(broadaddr->sin_len != sizeof(struct sockaddr_in) ||
broadaddr->sin_family != AF_INET))
return (EINVAL);
if (mask->sin_len != 0 &&
(mask->sin_len != sizeof(struct sockaddr_in) ||
mask->sin_family != AF_INET))
return (EINVAL);
if ((ifp->if_flags & IFF_POINTOPOINT) &&
(dstaddr->sin_len != sizeof(struct sockaddr_in) ||
dstaddr->sin_addr.s_addr == INADDR_ANY))
return (EDESTADDRREQ);
if (vhid != 0 && carp_attach_p == NULL)
return (EPROTONOSUPPORT);
/*
* See whether address already exist.
*/
iaIsFirst = true;
ia = NULL;
NET_EPOCH_ENTER(et);
CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
struct in_ifaddr *it;
if (ifa->ifa_addr->sa_family != AF_INET)
continue;
it = (struct in_ifaddr *)ifa;
if (it->ia_addr.sin_addr.s_addr == addr->sin_addr.s_addr &&
prison_check_ip4(td->td_ucred, &addr->sin_addr) == 0)
ia = it;
else
iaIsFirst = false;
}
NET_EPOCH_EXIT(et);
if (ia != NULL)
(void )in_difaddr_ioctl(cmd, data, ifp, td);
ifa = ifa_alloc(sizeof(struct in_ifaddr), M_WAITOK);
ia = (struct in_ifaddr *)ifa;
ifa->ifa_addr = (struct sockaddr *)&ia->ia_addr;
ifa->ifa_dstaddr = (struct sockaddr *)&ia->ia_dstaddr;
ifa->ifa_netmask = (struct sockaddr *)&ia->ia_sockmask;
callout_init_rw(&ia->ia_garp_timer, &ifp->if_addr_lock,
CALLOUT_RETURNUNLOCKED);
ia->ia_ifp = ifp;
ia->ia_addr = *addr;
if (mask->sin_len != 0) {
ia->ia_sockmask = *mask;
ia->ia_subnetmask = ntohl(ia->ia_sockmask.sin_addr.s_addr);
} else {
in_addr_t i = ntohl(addr->sin_addr.s_addr);
/*
* If netmask isn't supplied, use historical default.
* This is deprecated for interfaces other than loopback
* or point-to-point; warn in other cases. In the future
* we should return an error rather than warning.
*/
if ((ifp->if_flags & (IFF_POINTOPOINT | IFF_LOOPBACK)) == 0)
printf("%s: set address: WARNING: network mask "
"should be specified; using historical default\n",
ifp->if_xname);
if (IN_CLASSA(i))
ia->ia_subnetmask = IN_CLASSA_NET;
else if (IN_CLASSB(i))
ia->ia_subnetmask = IN_CLASSB_NET;
else
ia->ia_subnetmask = IN_CLASSC_NET;
ia->ia_sockmask.sin_addr.s_addr = htonl(ia->ia_subnetmask);
}
ia->ia_subnet = ntohl(addr->sin_addr.s_addr) & ia->ia_subnetmask;
in_socktrim(&ia->ia_sockmask);
if (ifp->if_flags & IFF_BROADCAST) {
if (broadaddr->sin_len != 0) {
ia->ia_broadaddr = *broadaddr;
} else if (ia->ia_subnetmask == IN_RFC3021_MASK) {
ia->ia_broadaddr.sin_addr.s_addr = INADDR_BROADCAST;
ia->ia_broadaddr.sin_len = sizeof(struct sockaddr_in);
ia->ia_broadaddr.sin_family = AF_INET;
} else {
ia->ia_broadaddr.sin_addr.s_addr =
htonl(ia->ia_subnet | ~ia->ia_subnetmask);
ia->ia_broadaddr.sin_len = sizeof(struct sockaddr_in);
ia->ia_broadaddr.sin_family = AF_INET;
}
}
if (ifp->if_flags & IFF_POINTOPOINT)
ia->ia_dstaddr = *dstaddr;
if (vhid != 0) {
error = (*carp_attach_p)(&ia->ia_ifa, vhid);
if (error)
return (error);
}
/* if_addrhead is already referenced by ifa_alloc() */
IF_ADDR_WLOCK(ifp);
CK_STAILQ_INSERT_TAIL(&ifp->if_addrhead, ifa, ifa_link);
IF_ADDR_WUNLOCK(ifp);
ifa_ref(ifa); /* in_ifaddrhead */
IN_IFADDR_WLOCK();
CK_STAILQ_INSERT_TAIL(&V_in_ifaddrhead, ia, ia_link);
LIST_INSERT_HEAD(INADDR_HASH(ia->ia_addr.sin_addr.s_addr), ia, ia_hash);
IN_IFADDR_WUNLOCK();
/*
* Give the interface a chance to initialize
* if this is its first address,
* and to validate the address if necessary.
*/
if (ifp->if_ioctl != NULL) {
error = (*ifp->if_ioctl)(ifp, SIOCSIFADDR, (caddr_t)ia);
if (error)
goto fail1;
}
/*
* Add route for the network.
*/
if (vhid == 0) {
error = in_addprefix(ia);
if (error)
goto fail1;
}
/*
* Add a loopback route to self.
*/
if (vhid == 0 && ia_need_loopback_route(ia)) {
struct in_ifaddr *eia;
eia = in_localip_more(ia);
if (eia == NULL) {
error = ifa_add_loopback_route((struct ifaddr *)ia,
(struct sockaddr *)&ia->ia_addr);
if (error)
goto fail2;
} else
ifa_free(&eia->ia_ifa);
}
if (iaIsFirst && (ifp->if_flags & IFF_MULTICAST)) {
struct in_addr allhosts_addr;
struct in_ifinfo *ii;
ii = ((struct in_ifinfo *)ifp->if_afdata[AF_INET]);
allhosts_addr.s_addr = htonl(INADDR_ALLHOSTS_GROUP);
error = in_joingroup(ifp, &allhosts_addr, NULL,
&ii->ii_allhosts);
}
/*
* Note: we don't need extra reference for ifa, since we called
* with sx lock held, and ifaddr can not be deleted in concurrent
* thread.
*/
EVENTHANDLER_INVOKE(ifaddr_event_ext, ifp, ifa, IFADDR_EVENT_ADD);
return (error);
fail2:
if (vhid == 0)
(void )in_scrubprefix(ia, LLE_STATIC);
fail1:
if (ia->ia_ifa.ifa_carp)
(*carp_detach_p)(&ia->ia_ifa, false);
IF_ADDR_WLOCK(ifp);
CK_STAILQ_REMOVE(&ifp->if_addrhead, &ia->ia_ifa, ifaddr, ifa_link);
IF_ADDR_WUNLOCK(ifp);
ifa_free(&ia->ia_ifa); /* if_addrhead */
IN_IFADDR_WLOCK();
CK_STAILQ_REMOVE(&V_in_ifaddrhead, ia, in_ifaddr, ia_link);
LIST_REMOVE(ia, ia_hash);
IN_IFADDR_WUNLOCK();
ifa_free(&ia->ia_ifa); /* in_ifaddrhead */
return (error);
}
static int
in_difaddr_ioctl(u_long cmd, caddr_t data, struct ifnet *ifp, struct thread *td)
{
const struct ifreq *ifr = (struct ifreq *)data;
const struct sockaddr_in *addr = (const struct sockaddr_in *)
&ifr->ifr_addr;
struct ifaddr *ifa;
struct in_ifaddr *ia;
bool deleteAny, iaIsLast;
int error;
if (td != NULL) {
error = priv_check(td, PRIV_NET_DELIFADDR);
if (error)
return (error);
}
if (addr->sin_len != sizeof(struct sockaddr_in) ||
addr->sin_family != AF_INET)
deleteAny = true;
else
deleteAny = false;
iaIsLast = true;
ia = NULL;
IF_ADDR_WLOCK(ifp);
CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
struct in_ifaddr *it;
if (ifa->ifa_addr->sa_family != AF_INET)
continue;
it = (struct in_ifaddr *)ifa;
if (deleteAny && ia == NULL && (td == NULL ||
prison_check_ip4(td->td_ucred, &it->ia_addr.sin_addr) == 0))
ia = it;
if (it->ia_addr.sin_addr.s_addr == addr->sin_addr.s_addr &&
(td == NULL || prison_check_ip4(td->td_ucred,
&addr->sin_addr) == 0))
ia = it;
if (it != ia)
iaIsLast = false;
}
if (ia == NULL) {
IF_ADDR_WUNLOCK(ifp);
return (EADDRNOTAVAIL);
}
CK_STAILQ_REMOVE(&ifp->if_addrhead, &ia->ia_ifa, ifaddr, ifa_link);
IF_ADDR_WUNLOCK(ifp);
ifa_free(&ia->ia_ifa); /* if_addrhead */
IN_IFADDR_WLOCK();
CK_STAILQ_REMOVE(&V_in_ifaddrhead, ia, in_ifaddr, ia_link);
LIST_REMOVE(ia, ia_hash);
IN_IFADDR_WUNLOCK();
/*
* in_scrubprefix() kills the interface route.
*/
in_scrubprefix(ia, LLE_STATIC);
/*
* in_ifadown gets rid of all the rest of
* the routes. This is not quite the right
* thing to do, but at least if we are running
* a routing process they will come back.
*/
in_ifadown(&ia->ia_ifa, 1);
if (ia->ia_ifa.ifa_carp)
(*carp_detach_p)(&ia->ia_ifa, cmd == SIOCAIFADDR);
/*
* If this is the last IPv4 address configured on this
* interface, leave the all-hosts group.
* No state-change report need be transmitted.
*/
if (iaIsLast && (ifp->if_flags & IFF_MULTICAST)) {
struct in_ifinfo *ii;
ii = ((struct in_ifinfo *)ifp->if_afdata[AF_INET]);
if (ii->ii_allhosts) {
(void)in_leavegroup(ii->ii_allhosts, NULL);
ii->ii_allhosts = NULL;
}
}
IF_ADDR_WLOCK(ifp);
if (callout_stop(&ia->ia_garp_timer) == 1) {
ifa_free(&ia->ia_ifa);
}
IF_ADDR_WUNLOCK(ifp);
EVENTHANDLER_INVOKE(ifaddr_event_ext, ifp, &ia->ia_ifa,
IFADDR_EVENT_DEL);
ifa_free(&ia->ia_ifa); /* in_ifaddrhead */
return (0);
}
static int
in_gifaddr_ioctl(u_long cmd, caddr_t data, struct ifnet *ifp, struct thread *td)
{
struct in_aliasreq *ifra = (struct in_aliasreq *)data;
const struct sockaddr_in *addr = &ifra->ifra_addr;
struct epoch_tracker et;
struct ifaddr *ifa;
struct in_ifaddr *ia;
/*
* ifra_addr must be present and be of INET family.
*/
if (addr->sin_len != sizeof(struct sockaddr_in) ||
addr->sin_family != AF_INET)
return (EINVAL);
/*
* See whether address exist.
*/
ia = NULL;
NET_EPOCH_ENTER(et);
CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
struct in_ifaddr *it;
if (ifa->ifa_addr->sa_family != AF_INET)
continue;
it = (struct in_ifaddr *)ifa;
if (it->ia_addr.sin_addr.s_addr == addr->sin_addr.s_addr &&
prison_check_ip4(td->td_ucred, &addr->sin_addr) == 0) {
ia = it;
break;
}
}
if (ia == NULL) {
NET_EPOCH_EXIT(et);
return (EADDRNOTAVAIL);
}
ifra->ifra_mask = ia->ia_sockmask;
if ((ifp->if_flags & IFF_POINTOPOINT) &&
ia->ia_dstaddr.sin_family == AF_INET)
ifra->ifra_dstaddr = ia->ia_dstaddr;
else if ((ifp->if_flags & IFF_BROADCAST) &&
ia->ia_broadaddr.sin_family == AF_INET)
ifra->ifra_broadaddr = ia->ia_broadaddr;
else
memset(&ifra->ifra_broadaddr, 0,
sizeof(ifra->ifra_broadaddr));
NET_EPOCH_EXIT(et);
return (0);
}
static int
in_match_ifaddr(const struct rtentry *rt, const struct nhop_object *nh, void *arg)
{
if (nh->nh_ifa == (struct ifaddr *)arg)
return (1);
return (0);
}
static int
in_handle_prefix_route(uint32_t fibnum, int cmd,
struct sockaddr_in *dst, struct sockaddr_in *netmask, struct ifaddr *ifa,
struct ifnet *ifp)
{
NET_EPOCH_ASSERT();
/* Prepare gateway */
struct sockaddr_dl_short sdl = {
.sdl_family = AF_LINK,
.sdl_len = sizeof(struct sockaddr_dl_short),
.sdl_type = ifa->ifa_ifp->if_type,
.sdl_index = ifa->ifa_ifp->if_index,
};
struct rt_addrinfo info = {
.rti_ifa = ifa,
.rti_ifp = ifp,
.rti_flags = RTF_PINNED | ((netmask != NULL) ? 0 : RTF_HOST),
.rti_info = {
[RTAX_DST] = (struct sockaddr *)dst,
[RTAX_NETMASK] = (struct sockaddr *)netmask,
[RTAX_GATEWAY] = (struct sockaddr *)&sdl,
},
/* Ensure we delete the prefix IFF prefix ifa matches */
.rti_filter = in_match_ifaddr,
.rti_filterdata = ifa,
};
return (rib_handle_ifaddr_info(fibnum, cmd, &info));
}
/*
* Routing table interaction with interface addresses.
*
* In general, two types of routes needs to be installed:
* a) "interface" or "prefix" route, telling user that the addresses
* behind the ifa prefix are reached directly.
* b) "loopback" route installed for the ifa address, telling user that
* the address belongs to local system.
*
* Handling for (a) and (b) differs in multi-fib aspects, hence they
* are implemented in different functions below.
*
* The cases above may intersect - /32 interface aliases results in
* the same prefix produced by (a) and (b). This blurs the definition
* of the "loopback" route and complicate interactions. The interaction
* table is defined below. The case numbers are used in the multiple
* functions below to refer to the particular test case.
*
* There can be multiple options:
* 1) Adding address with prefix on non-p2p/non-loopback interface.
* Example: 192.0.2.1/24. Action:
* * add "prefix" route towards 192.0.2.0/24 via @ia interface,
* using @ia as an address source.
* * add "loopback" route towards 192.0.2.1 via V_loif, saving
* @ia ifp in the gateway and using @ia as an address source.
*
* 2) Adding address with /32 mask to non-p2p/non-loopback interface.
* Example: 192.0.2.2/32. Action:
* * add "prefix" host route via V_loif, using @ia as an address source.
*
* 3) Adding address with or without prefix to p2p interface.
* Example: 10.0.0.1/24->10.0.0.2. Action:
* * add "prefix" host route towards 10.0.0.2 via this interface, using @ia
* as an address source. Note: no sense in installing full /24 as the interface
* is point-to-point.
* * add "loopback" route towards 10.0.9.1 via V_loif, saving
* @ia ifp in the gateway and using @ia as an address source.
*
* 4) Adding address with or without prefix to loopback interface.
* Example: 192.0.2.1/24. Action:
* * add "prefix" host route via @ia interface, using @ia as an address source.
* Note: Skip installing /24 prefix as it would introduce TTL loop
* for the traffic destined to these addresses.
*/
/*
* Checks if @ia needs to install loopback route to @ia address via
* ifa_maintain_loopback_route().
*
* Return true on success.
*/
static bool
ia_need_loopback_route(const struct in_ifaddr *ia)
{
struct ifnet *ifp = ia->ia_ifp;
/* Case 4: Skip loopback interfaces */
if ((ifp->if_flags & IFF_LOOPBACK) ||
(ia->ia_addr.sin_addr.s_addr == INADDR_ANY))
return (false);
/* Clash avoidance: Skip p2p interfaces with both addresses are equal */
if ((ifp->if_flags & IFF_POINTOPOINT) &&
ia->ia_dstaddr.sin_addr.s_addr == ia->ia_addr.sin_addr.s_addr)
return (false);
/* Case 2: skip /32 prefixes */
if (!(ifp->if_flags & IFF_POINTOPOINT) &&
(ia->ia_sockmask.sin_addr.s_addr == INADDR_BROADCAST))
return (false);
return (true);
}
/*
* Calculate "prefix" route corresponding to @ia.
*/
static void
ia_getrtprefix(const struct in_ifaddr *ia, struct in_addr *prefix, struct in_addr *mask)
{
if (ia->ia_ifp->if_flags & IFF_POINTOPOINT) {
/* Case 3: return host route for dstaddr */
*prefix = ia->ia_dstaddr.sin_addr;
mask->s_addr = INADDR_BROADCAST;
} else if (ia->ia_ifp->if_flags & IFF_LOOPBACK) {
/* Case 4: return host route for ifaddr */
*prefix = ia->ia_addr.sin_addr;
mask->s_addr = INADDR_BROADCAST;
} else {
/* Cases 1,2: return actual ia prefix */
*prefix = ia->ia_addr.sin_addr;
*mask = ia->ia_sockmask.sin_addr;
prefix->s_addr &= mask->s_addr;
}
}
/*
* Adds or delete interface "prefix" route corresponding to @ifa.
* Returns 0 on success or errno.
*/
int
in_handle_ifaddr_route(int cmd, struct in_ifaddr *ia)
{
struct ifaddr *ifa = &ia->ia_ifa;
struct in_addr daddr, maddr;
struct sockaddr_in *pmask;
struct epoch_tracker et;
int error;
ia_getrtprefix(ia, &daddr, &maddr);
struct sockaddr_in mask = {
.sin_family = AF_INET,
.sin_len = sizeof(struct sockaddr_in),
.sin_addr = maddr,
};
pmask = (maddr.s_addr != INADDR_BROADCAST) ? &mask : NULL;
struct sockaddr_in dst = {
.sin_family = AF_INET,
.sin_len = sizeof(struct sockaddr_in),
.sin_addr.s_addr = daddr.s_addr & maddr.s_addr,
};
struct ifnet *ifp = ia->ia_ifp;
if ((maddr.s_addr == INADDR_BROADCAST) &&
(!(ia->ia_ifp->if_flags & (IFF_POINTOPOINT|IFF_LOOPBACK)))) {
/* Case 2: host route on broadcast interface */
ifp = V_loif;
}
uint32_t fibnum = ifa->ifa_ifp->if_fib;
NET_EPOCH_ENTER(et);
error = in_handle_prefix_route(fibnum, cmd, &dst, pmask, ifa, ifp);
NET_EPOCH_EXIT(et);
return (error);
}
/*
* Check if we have a route for the given prefix already.
*/
static bool
in_hasrtprefix(struct in_ifaddr *target)
{
struct rm_priotracker in_ifa_tracker;
struct in_ifaddr *ia;
struct in_addr prefix, mask, p, m;
bool result = false;
ia_getrtprefix(target, &prefix, &mask);
IN_IFADDR_RLOCK(&in_ifa_tracker);
/* Look for an existing address with the same prefix, mask, and fib */
CK_STAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
ia_getrtprefix(ia, &p, &m);
if (prefix.s_addr != p.s_addr ||
mask.s_addr != m.s_addr)
continue;
if (target->ia_ifp->if_fib != ia->ia_ifp->if_fib)
continue;
/*
* If we got a matching prefix route inserted by other
* interface address, we are done here.
*/
if (ia->ia_flags & IFA_ROUTE) {
result = true;
break;
}
}
IN_IFADDR_RUNLOCK(&in_ifa_tracker);
return (result);
}
int
in_addprefix(struct in_ifaddr *target)
{
int error;
if (in_hasrtprefix(target)) {
if (V_nosameprefix)
return (EEXIST);
else {
rt_addrmsg(RTM_ADD, &target->ia_ifa,
target->ia_ifp->if_fib);
return (0);
}
}
/*
* No-one seem to have this prefix route, so we try to insert it.
*/
rt_addrmsg(RTM_ADD, &target->ia_ifa, target->ia_ifp->if_fib);
error = in_handle_ifaddr_route(RTM_ADD, target);
if (!error)
target->ia_flags |= IFA_ROUTE;
return (error);
}
/*
* Removes either all lle entries for given @ia, or lle
* corresponding to @ia address.
*/
static void
in_scrubprefixlle(struct in_ifaddr *ia, int all, u_int flags)
{
struct sockaddr_in addr, mask;
struct sockaddr *saddr, *smask;
struct ifnet *ifp;
saddr = (struct sockaddr *)&addr;
bzero(&addr, sizeof(addr));
addr.sin_len = sizeof(addr);
addr.sin_family = AF_INET;
smask = (struct sockaddr *)&mask;
bzero(&mask, sizeof(mask));
mask.sin_len = sizeof(mask);
mask.sin_family = AF_INET;
mask.sin_addr.s_addr = ia->ia_subnetmask;
ifp = ia->ia_ifp;
if (all) {
/*
* Remove all L2 entries matching given prefix.
* Convert address to host representation to avoid
* doing this on every callback. ia_subnetmask is already
* stored in host representation.
*/
addr.sin_addr.s_addr = ntohl(ia->ia_addr.sin_addr.s_addr);
lltable_prefix_free(AF_INET, saddr, smask, flags);
} else {
/* Remove interface address only */
addr.sin_addr.s_addr = ia->ia_addr.sin_addr.s_addr;
lltable_delete_addr(LLTABLE(ifp), LLE_IFADDR, saddr);
}
}
/*
* If there is no other address in the system that can serve a route to the
* same prefix, remove the route. Hand over the route to the new address
* otherwise.
*/
int
in_scrubprefix(struct in_ifaddr *target, u_int flags)
{
struct rm_priotracker in_ifa_tracker;
struct in_ifaddr *ia;
struct in_addr prefix, mask, p, m;
int error = 0;
/*
* Remove the loopback route to the interface address.
*/
if (ia_need_loopback_route(target) && (flags & LLE_STATIC)) {
struct in_ifaddr *eia;
eia = in_localip_more(target);
if (eia != NULL) {
error = ifa_switch_loopback_route((struct ifaddr *)eia,
(struct sockaddr *)&target->ia_addr);
ifa_free(&eia->ia_ifa);
} else {
error = ifa_del_loopback_route((struct ifaddr *)target,
(struct sockaddr *)&target->ia_addr);
}
}
ia_getrtprefix(target, &prefix, &mask);
if ((target->ia_flags & IFA_ROUTE) == 0) {
rt_addrmsg(RTM_DELETE, &target->ia_ifa, target->ia_ifp->if_fib);
/*
* Removing address from !IFF_UP interface or
* prefix which exists on other interface (along with route).
* No entries should exist here except target addr.
* Given that, delete this entry only.
*/
in_scrubprefixlle(target, 0, flags);
return (0);
}
IN_IFADDR_RLOCK(&in_ifa_tracker);
CK_STAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
ia_getrtprefix(ia, &p, &m);
if (prefix.s_addr != p.s_addr ||
mask.s_addr != m.s_addr)
continue;
if ((ia->ia_ifp->if_flags & IFF_UP) == 0)
continue;
/*
* If we got a matching prefix address, move IFA_ROUTE and
* the route itself to it. Make sure that routing daemons
* get a heads-up.
*/
if ((ia->ia_flags & IFA_ROUTE) == 0) {
ifa_ref(&ia->ia_ifa);
IN_IFADDR_RUNLOCK(&in_ifa_tracker);
error = in_handle_ifaddr_route(RTM_DELETE, target);
if (error == 0)
target->ia_flags &= ~IFA_ROUTE;
else
log(LOG_INFO, "in_scrubprefix: err=%d, old prefix delete failed\n",
error);
/* Scrub all entries IFF interface is different */
in_scrubprefixlle(target, target->ia_ifp != ia->ia_ifp,
flags);
error = in_handle_ifaddr_route(RTM_ADD, ia);
if (error == 0)
ia->ia_flags |= IFA_ROUTE;
else
log(LOG_INFO, "in_scrubprefix: err=%d, new prefix add failed\n",
error);
ifa_free(&ia->ia_ifa);
return (error);
}
}
IN_IFADDR_RUNLOCK(&in_ifa_tracker);
/*
* remove all L2 entries on the given prefix
*/
in_scrubprefixlle(target, 1, flags);
/*
* As no-one seem to have this prefix, we can remove the route.
*/
rt_addrmsg(RTM_DELETE, &target->ia_ifa, target->ia_ifp->if_fib);
error = in_handle_ifaddr_route(RTM_DELETE, target);
if (error == 0)
target->ia_flags &= ~IFA_ROUTE;
else
log(LOG_INFO, "in_scrubprefix: err=%d, prefix delete failed\n", error);
return (error);
}
void
in_ifscrub_all(void)
{
struct ifnet *ifp;
struct ifaddr *ifa, *nifa;
struct ifaliasreq ifr;
IFNET_RLOCK();
CK_STAILQ_FOREACH(ifp, &V_ifnet, if_link) {
/* Cannot lock here - lock recursion. */
/* NET_EPOCH_ENTER(et); */
CK_STAILQ_FOREACH_SAFE(ifa, &ifp->if_addrhead, ifa_link, nifa) {
if (ifa->ifa_addr->sa_family != AF_INET)
continue;
/*
* This is ugly but the only way for legacy IP to
* cleanly remove addresses and everything attached.
*/
bzero(&ifr, sizeof(ifr));
ifr.ifra_addr = *ifa->ifa_addr;
if (ifa->ifa_dstaddr)
ifr.ifra_broadaddr = *ifa->ifa_dstaddr;
(void)in_control(NULL, SIOCDIFADDR, (caddr_t)&ifr,
ifp, NULL);
}
/* NET_EPOCH_EXIT(et); */
in_purgemaddrs(ifp);
igmp_domifdetach(ifp);
}
IFNET_RUNLOCK();
}
int
in_ifaddr_broadcast(struct in_addr in, struct in_ifaddr *ia)
{
return ((in.s_addr == ia->ia_broadaddr.sin_addr.s_addr ||
/*
* Optionally check for old-style (host 0) broadcast, but
* taking into account that RFC 3021 obsoletes it.
*/
(V_broadcast_lowest && ia->ia_subnetmask != IN_RFC3021_MASK &&
ntohl(in.s_addr) == ia->ia_subnet)) &&
/*
* Check for an all one subnetmask. These
* only exist when an interface gets a secondary
* address.
*/
ia->ia_subnetmask != (u_long)0xffffffff);
}
/*
* Return 1 if the address might be a local broadcast address.
*/
int
in_broadcast(struct in_addr in, struct ifnet *ifp)
{
struct ifaddr *ifa;
int found;
NET_EPOCH_ASSERT();
if (in.s_addr == INADDR_BROADCAST ||
in.s_addr == INADDR_ANY)
return (1);
if ((ifp->if_flags & IFF_BROADCAST) == 0)
return (0);
found = 0;
/*
* Look through the list of addresses for a match
* with a broadcast address.
*/
CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link)
if (ifa->ifa_addr->sa_family == AF_INET &&
in_ifaddr_broadcast(in, (struct in_ifaddr *)ifa)) {
found = 1;
break;
}
return (found);
}
/*
* On interface removal, clean up IPv4 data structures hung off of the ifnet.
*/
void
in_ifdetach(struct ifnet *ifp)
{
IN_MULTI_LOCK();
in_pcbpurgeif0(&V_ripcbinfo, ifp);
in_pcbpurgeif0(&V_udbinfo, ifp);
in_pcbpurgeif0(&V_ulitecbinfo, ifp);
in_purgemaddrs(ifp);
IN_MULTI_UNLOCK();
/*
* Make sure all multicast deletions invoking if_ioctl() are
* completed before returning. Else we risk accessing a freed
* ifnet structure pointer.
*/
inm_release_wait(NULL);
}
/*
* Delete all IPv4 multicast address records, and associated link-layer
* multicast address records, associated with ifp.
* XXX It looks like domifdetach runs AFTER the link layer cleanup.
* XXX This should not race with ifma_protospec being set during
* a new allocation, if it does, we have bigger problems.
*/
static void
in_purgemaddrs(struct ifnet *ifp)
{
struct in_multi_head purgeinms;
struct in_multi *inm;
struct ifmultiaddr *ifma, *next;
SLIST_INIT(&purgeinms);
IN_MULTI_LIST_LOCK();
/*
* Extract list of in_multi associated with the detaching ifp
* which the PF_INET layer is about to release.
* We need to do this as IF_ADDR_LOCK() may be re-acquired
* by code further down.
*/
IF_ADDR_WLOCK(ifp);
restart:
CK_STAILQ_FOREACH_SAFE(ifma, &ifp->if_multiaddrs, ifma_link, next) {
if (ifma->ifma_addr->sa_family != AF_INET ||
ifma->ifma_protospec == NULL)
continue;
inm = (struct in_multi *)ifma->ifma_protospec;
inm_rele_locked(&purgeinms, inm);
if (__predict_false(ifma_restart)) {
ifma_restart = true;
goto restart;
}
}
IF_ADDR_WUNLOCK(ifp);
inm_release_list_deferred(&purgeinms);
igmp_ifdetach(ifp);
IN_MULTI_LIST_UNLOCK();
}
struct in_llentry {
struct llentry base;
};
#define IN_LLTBL_DEFAULT_HSIZE 32
#define IN_LLTBL_HASH(k, h) \
(((((((k >> 8) ^ k) >> 8) ^ k) >> 8) ^ k) & ((h) - 1))
/*
* Do actual deallocation of @lle.
*/
static void
in_lltable_destroy_lle_unlocked(epoch_context_t ctx)
{
struct llentry *lle;
lle = __containerof(ctx, struct llentry, lle_epoch_ctx);
LLE_LOCK_DESTROY(lle);
LLE_REQ_DESTROY(lle);
free(lle, M_LLTABLE);
}
/*
* Called by LLE_FREE_LOCKED when number of references
* drops to zero.
*/
static void
in_lltable_destroy_lle(struct llentry *lle)
{
LLE_WUNLOCK(lle);
NET_EPOCH_CALL(in_lltable_destroy_lle_unlocked, &lle->lle_epoch_ctx);
}
static struct llentry *
in_lltable_new(struct in_addr addr4, u_int flags)
{
struct in_llentry *lle;
lle = malloc(sizeof(struct in_llentry), M_LLTABLE, M_NOWAIT | M_ZERO);
if (lle == NULL) /* NB: caller generates msg */
return NULL;
/*
* For IPv4 this will trigger "arpresolve" to generate
* an ARP request.
*/
lle->base.la_expire = time_uptime; /* mark expired */
lle->base.r_l3addr.addr4 = addr4;
lle->base.lle_refcnt = 1;
lle->base.lle_free = in_lltable_destroy_lle;
LLE_LOCK_INIT(&lle->base);
LLE_REQ_INIT(&lle->base);
callout_init(&lle->base.lle_timer, 1);
return (&lle->base);
}
#define IN_ARE_MASKED_ADDR_EQUAL(d, a, m) ( \
((((d).s_addr ^ (a).s_addr) & (m).s_addr)) == 0 )
static int
in_lltable_match_prefix(const struct sockaddr *saddr,
const struct sockaddr *smask, u_int flags, struct llentry *lle)
{
struct in_addr addr, mask, lle_addr;
addr = ((const struct sockaddr_in *)saddr)->sin_addr;
mask = ((const struct sockaddr_in *)smask)->sin_addr;
lle_addr.s_addr = ntohl(lle->r_l3addr.addr4.s_addr);
if (IN_ARE_MASKED_ADDR_EQUAL(lle_addr, addr, mask) == 0)
return (0);
if (lle->la_flags & LLE_IFADDR) {
/*
* Delete LLE_IFADDR records IFF address & flag matches.
* Note that addr is the interface address within prefix
* being matched.
* Note also we should handle 'ifdown' cases without removing
* ifaddr macs.
*/
if (addr.s_addr == lle_addr.s_addr && (flags & LLE_STATIC) != 0)
return (1);
return (0);
}
/* flags & LLE_STATIC means deleting both dynamic and static entries */
if ((flags & LLE_STATIC) || !(lle->la_flags & LLE_STATIC))
return (1);
return (0);
}
static void
in_lltable_free_entry(struct lltable *llt, struct llentry *lle)
{
size_t pkts_dropped;
LLE_WLOCK_ASSERT(lle);
KASSERT(llt != NULL, ("lltable is NULL"));
/* Unlink entry from table if not already */
if ((lle->la_flags & LLE_LINKED) != 0) {
IF_AFDATA_WLOCK_ASSERT(llt->llt_ifp);
lltable_unlink_entry(llt, lle);
}
/* Drop hold queue */
pkts_dropped = llentry_free(lle);
ARPSTAT_ADD(dropped, pkts_dropped);
}
static int
in_lltable_rtcheck(struct ifnet *ifp, u_int flags, const struct sockaddr *l3addr)
{
struct nhop_object *nh;
struct in_addr addr;
KASSERT(l3addr->sa_family == AF_INET,
("sin_family %d", l3addr->sa_family));
addr = ((const struct sockaddr_in *)l3addr)->sin_addr;
nh = fib4_lookup(ifp->if_fib, addr, 0, NHR_NONE, 0);
if (nh == NULL)
return (EINVAL);
/*
* If the gateway for an existing host route matches the target L3
* address, which is a special route inserted by some implementation
* such as MANET, and the interface is of the correct type, then
* allow for ARP to proceed.
*/
if (nh->nh_flags & NHF_GATEWAY) {
if (!(nh->nh_flags & NHF_HOST) || nh->nh_ifp->if_type != IFT_ETHER ||
(nh->nh_ifp->if_flags & (IFF_NOARP | IFF_STATICARP)) != 0 ||
memcmp(nh->gw_sa.sa_data, l3addr->sa_data,
sizeof(in_addr_t)) != 0) {
return (EINVAL);
}
}
/*
* Make sure that at least the destination address is covered
* by the route. This is for handling the case where 2 or more
* interfaces have the same prefix. An incoming packet arrives
* on one interface and the corresponding outgoing packet leaves
* another interface.
*/
if ((nh->nh_ifp != ifp) && (nh->nh_flags & NHF_HOST) == 0) {
struct in_ifaddr *ia = (struct in_ifaddr *)ifaof_ifpforaddr(l3addr, ifp);
struct in_addr dst_addr, mask_addr;
if (ia == NULL)
return (EINVAL);
/*
* ifaof_ifpforaddr() returns _best matching_ IFA.
* It is possible that ifa prefix does not cover our address.
* Explicitly verify and fail if that's the case.
*/
dst_addr = IA_SIN(ia)->sin_addr;
mask_addr.s_addr = htonl(ia->ia_subnetmask);
if (!IN_ARE_MASKED_ADDR_EQUAL(dst_addr, addr, mask_addr))
return (EINVAL);
}
return (0);
}
static inline uint32_t
in_lltable_hash_dst(const struct in_addr dst, uint32_t hsize)
{
return (IN_LLTBL_HASH(dst.s_addr, hsize));
}
static uint32_t
in_lltable_hash(const struct llentry *lle, uint32_t hsize)
{
return (in_lltable_hash_dst(lle->r_l3addr.addr4, hsize));
}
static void
in_lltable_fill_sa_entry(const struct llentry *lle, struct sockaddr *sa)
{
struct sockaddr_in *sin;
sin = (struct sockaddr_in *)sa;
bzero(sin, sizeof(*sin));
sin->sin_family = AF_INET;
sin->sin_len = sizeof(*sin);
sin->sin_addr = lle->r_l3addr.addr4;
}
static inline struct llentry *
in_lltable_find_dst(struct lltable *llt, struct in_addr dst)
{
struct llentry *lle;
struct llentries *lleh;
u_int hashidx;
hashidx = in_lltable_hash_dst(dst, llt->llt_hsize);
lleh = &llt->lle_head[hashidx];
CK_LIST_FOREACH(lle, lleh, lle_next) {
if (lle->la_flags & LLE_DELETED)
continue;
if (lle->r_l3addr.addr4.s_addr == dst.s_addr)
break;
}
return (lle);
}
static void
in_lltable_delete_entry(struct lltable *llt, struct llentry *lle)
{
lle->la_flags |= LLE_DELETED;
EVENTHANDLER_INVOKE(lle_event, lle, LLENTRY_DELETED);
#ifdef DIAGNOSTIC
log(LOG_INFO, "ifaddr cache = %p is deleted\n", lle);
#endif
llentry_free(lle);
}
static struct llentry *
in_lltable_alloc(struct lltable *llt, u_int flags, const struct sockaddr *l3addr)
{
const struct sockaddr_in *sin = (const struct sockaddr_in *)l3addr;
struct ifnet *ifp = llt->llt_ifp;
struct llentry *lle;
char linkhdr[LLE_MAX_LINKHDR];
size_t linkhdrsize;
int lladdr_off;
KASSERT(l3addr->sa_family == AF_INET,
("sin_family %d", l3addr->sa_family));
/*
* A route that covers the given address must have
* been installed 1st because we are doing a resolution,
* verify this.
*/
if (!(flags & LLE_IFADDR) &&
in_lltable_rtcheck(ifp, flags, l3addr) != 0)
return (NULL);
lle = in_lltable_new(sin->sin_addr, flags);
if (lle == NULL) {
log(LOG_INFO, "lla_lookup: new lle malloc failed\n");
return (NULL);
}
lle->la_flags = flags;
if (flags & LLE_STATIC)
lle->r_flags |= RLLE_VALID;
if ((flags & LLE_IFADDR) == LLE_IFADDR) {
linkhdrsize = LLE_MAX_LINKHDR;
if (lltable_calc_llheader(ifp, AF_INET, IF_LLADDR(ifp),
linkhdr, &linkhdrsize, &lladdr_off) != 0) {
- NET_EPOCH_CALL(in_lltable_destroy_lle_unlocked, &lle->lle_epoch_ctx);
+ in_lltable_free_entry(llt, lle);
return (NULL);
}
lltable_set_entry_addr(ifp, lle, linkhdr, linkhdrsize,
lladdr_off);
lle->la_flags |= LLE_STATIC;
lle->r_flags |= (RLLE_VALID | RLLE_IFADDR);
}
return (lle);
}
/*
* Return NULL if not found or marked for deletion.
* If found return lle read locked.
*/
static struct llentry *
in_lltable_lookup(struct lltable *llt, u_int flags, const struct sockaddr *l3addr)
{
const struct sockaddr_in *sin = (const struct sockaddr_in *)l3addr;
struct llentry *lle;
IF_AFDATA_LOCK_ASSERT(llt->llt_ifp);
KASSERT(l3addr->sa_family == AF_INET,
("sin_family %d", l3addr->sa_family));
KASSERT((flags & (LLE_UNLOCKED | LLE_EXCLUSIVE)) !=
(LLE_UNLOCKED | LLE_EXCLUSIVE),
("wrong lle request flags: %#x", flags));
lle = in_lltable_find_dst(llt, sin->sin_addr);
if (lle == NULL)
return (NULL);
if (flags & LLE_UNLOCKED)
return (lle);
if (flags & LLE_EXCLUSIVE)
LLE_WLOCK(lle);
else
LLE_RLOCK(lle);
/*
* If the afdata lock is not held, the LLE may have been unlinked while
* we were blocked on the LLE lock. Check for this case.
*/
if (__predict_false((lle->la_flags & LLE_LINKED) == 0)) {
if (flags & LLE_EXCLUSIVE)
LLE_WUNLOCK(lle);
else
LLE_RUNLOCK(lle);
return (NULL);
}
return (lle);
}
static int
in_lltable_dump_entry(struct lltable *llt, struct llentry *lle,
struct sysctl_req *wr)
{
struct ifnet *ifp = llt->llt_ifp;
/* XXX stack use */
struct {
struct rt_msghdr rtm;
struct sockaddr_in sin;
struct sockaddr_dl sdl;
} arpc;
struct sockaddr_dl *sdl;
int error;
bzero(&arpc, sizeof(arpc));
/* skip deleted entries */
if ((lle->la_flags & LLE_DELETED) == LLE_DELETED)
return (0);
/* Skip if jailed and not a valid IP of the prison. */
lltable_fill_sa_entry(lle,(struct sockaddr *)&arpc.sin);
if (prison_if(wr->td->td_ucred, (struct sockaddr *)&arpc.sin) != 0)
return (0);
/*
* produce a msg made of:
* struct rt_msghdr;
* struct sockaddr_in; (IPv4)
* struct sockaddr_dl;
*/
arpc.rtm.rtm_msglen = sizeof(arpc);
arpc.rtm.rtm_version = RTM_VERSION;
arpc.rtm.rtm_type = RTM_GET;
arpc.rtm.rtm_flags = RTF_UP;
arpc.rtm.rtm_addrs = RTA_DST | RTA_GATEWAY;
/* publish */
if (lle->la_flags & LLE_PUB)
arpc.rtm.rtm_flags |= RTF_ANNOUNCE;
sdl = &arpc.sdl;
sdl->sdl_family = AF_LINK;
sdl->sdl_len = sizeof(*sdl);
sdl->sdl_index = ifp->if_index;
sdl->sdl_type = ifp->if_type;
if ((lle->la_flags & LLE_VALID) == LLE_VALID) {
sdl->sdl_alen = ifp->if_addrlen;
bcopy(lle->ll_addr, LLADDR(sdl), ifp->if_addrlen);
} else {
sdl->sdl_alen = 0;
bzero(LLADDR(sdl), ifp->if_addrlen);
}
arpc.rtm.rtm_rmx.rmx_expire =
lle->la_flags & LLE_STATIC ? 0 : lle->la_expire;
arpc.rtm.rtm_flags |= (RTF_HOST | RTF_LLDATA);
if (lle->la_flags & LLE_STATIC)
arpc.rtm.rtm_flags |= RTF_STATIC;
if (lle->la_flags & LLE_IFADDR)
arpc.rtm.rtm_flags |= RTF_PINNED;
arpc.rtm.rtm_index = ifp->if_index;
error = SYSCTL_OUT(wr, &arpc, sizeof(arpc));
return (error);
}
static struct lltable *
in_lltattach(struct ifnet *ifp)
{
struct lltable *llt;
llt = lltable_allocate_htbl(IN_LLTBL_DEFAULT_HSIZE);
llt->llt_af = AF_INET;
llt->llt_ifp = ifp;
llt->llt_lookup = in_lltable_lookup;
llt->llt_alloc_entry = in_lltable_alloc;
llt->llt_delete_entry = in_lltable_delete_entry;
llt->llt_dump_entry = in_lltable_dump_entry;
llt->llt_hash = in_lltable_hash;
llt->llt_fill_sa_entry = in_lltable_fill_sa_entry;
llt->llt_free_entry = in_lltable_free_entry;
llt->llt_match_prefix = in_lltable_match_prefix;
llt->llt_mark_used = llentry_mark_used;
lltable_link(llt);
return (llt);
}
struct lltable *
in_lltable_get(struct ifnet *ifp)
{
struct lltable *llt = NULL;
void *afdata_ptr = ifp->if_afdata[AF_INET];
if (afdata_ptr != NULL)
llt = ((struct in_ifinfo *)afdata_ptr)->ii_llt;
return (llt);
}
void *
in_domifattach(struct ifnet *ifp)
{
struct in_ifinfo *ii;
ii = malloc(sizeof(struct in_ifinfo), M_IFADDR, M_WAITOK|M_ZERO);
ii->ii_llt = in_lltattach(ifp);
ii->ii_igmp = igmp_domifattach(ifp);
return (ii);
}
void
in_domifdetach(struct ifnet *ifp, void *aux)
{
struct in_ifinfo *ii = (struct in_ifinfo *)aux;
igmp_domifdetach(ifp);
lltable_free(ii->ii_llt);
free(ii, M_IFADDR);
}
diff --git a/sys/netinet6/in6.c b/sys/netinet6/in6.c
index 41b05ae2ca53..b0a8c1608b6c 100644
--- a/sys/netinet6/in6.c
+++ b/sys/netinet6/in6.c
@@ -1,2595 +1,2595 @@
/*-
* SPDX-License-Identifier: BSD-3-Clause
*
* Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the project nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* $KAME: in6.c,v 1.259 2002/01/21 11:37:50 keiichi Exp $
*/
/*-
* Copyright (c) 1982, 1986, 1991, 1993
* The Regents of the University of California. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* @(#)in.c 8.2 (Berkeley) 11/15/93
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include "opt_inet.h"
#include "opt_inet6.h"
#include <sys/param.h>
#include <sys/eventhandler.h>
#include <sys/errno.h>
#include <sys/jail.h>
#include <sys/malloc.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/sockio.h>
#include <sys/systm.h>
#include <sys/priv.h>
#include <sys/proc.h>
#include <sys/protosw.h>
#include <sys/time.h>
#include <sys/kernel.h>
#include <sys/lock.h>
#include <sys/rmlock.h>
#include <sys/sysctl.h>
#include <sys/syslog.h>
#include <net/if.h>
#include <net/if_var.h>
#include <net/if_types.h>
#include <net/route.h>
#include <net/route/route_ctl.h>
#include <net/route/nhop.h>
#include <net/if_dl.h>
#include <net/vnet.h>
#include <netinet/in.h>
#include <netinet/in_var.h>
#include <net/if_llatbl.h>
#include <netinet/if_ether.h>
#include <netinet/in_systm.h>
#include <netinet/ip.h>
#include <netinet/in_pcb.h>
#include <netinet/ip_carp.h>
#include <netinet/ip6.h>
#include <netinet6/ip6_var.h>
#include <netinet6/nd6.h>
#include <netinet6/mld6_var.h>
#include <netinet6/ip6_mroute.h>
#include <netinet6/in6_ifattach.h>
#include <netinet6/scope6_var.h>
#include <netinet6/in6_fib.h>
#include <netinet6/in6_pcb.h>
/*
* struct in6_ifreq and struct ifreq must be type punnable for common members
* of ifr_ifru to allow accessors to be shared.
*/
_Static_assert(offsetof(struct in6_ifreq, ifr_ifru) ==
offsetof(struct ifreq, ifr_ifru),
"struct in6_ifreq and struct ifreq are not type punnable");
VNET_DECLARE(int, icmp6_nodeinfo_oldmcprefix);
#define V_icmp6_nodeinfo_oldmcprefix VNET(icmp6_nodeinfo_oldmcprefix)
/*
* Definitions of some costant IP6 addresses.
*/
const struct in6_addr in6addr_any = IN6ADDR_ANY_INIT;
const struct in6_addr in6addr_loopback = IN6ADDR_LOOPBACK_INIT;
const struct in6_addr in6addr_nodelocal_allnodes =
IN6ADDR_NODELOCAL_ALLNODES_INIT;
const struct in6_addr in6addr_linklocal_allnodes =
IN6ADDR_LINKLOCAL_ALLNODES_INIT;
const struct in6_addr in6addr_linklocal_allrouters =
IN6ADDR_LINKLOCAL_ALLROUTERS_INIT;
const struct in6_addr in6addr_linklocal_allv2routers =
IN6ADDR_LINKLOCAL_ALLV2ROUTERS_INIT;
const struct in6_addr in6mask0 = IN6MASK0;
const struct in6_addr in6mask32 = IN6MASK32;
const struct in6_addr in6mask64 = IN6MASK64;
const struct in6_addr in6mask96 = IN6MASK96;
const struct in6_addr in6mask128 = IN6MASK128;
const struct sockaddr_in6 sa6_any =
{ sizeof(sa6_any), AF_INET6, 0, 0, IN6ADDR_ANY_INIT, 0 };
static int in6_notify_ifa(struct ifnet *, struct in6_ifaddr *,
struct in6_aliasreq *, int);
static void in6_unlink_ifa(struct in6_ifaddr *, struct ifnet *);
static int in6_validate_ifra(struct ifnet *, struct in6_aliasreq *,
struct in6_ifaddr *, int);
static struct in6_ifaddr *in6_alloc_ifa(struct ifnet *,
struct in6_aliasreq *, int flags);
static int in6_update_ifa_internal(struct ifnet *, struct in6_aliasreq *,
struct in6_ifaddr *, int, int);
static int in6_broadcast_ifa(struct ifnet *, struct in6_aliasreq *,
struct in6_ifaddr *, int);
#define ifa2ia6(ifa) ((struct in6_ifaddr *)(ifa))
#define ia62ifa(ia6) (&((ia6)->ia_ifa))
void
in6_newaddrmsg(struct in6_ifaddr *ia, int cmd)
{
struct rt_addrinfo info;
struct ifaddr *ifa;
struct sockaddr_dl gateway;
int fibnum;
ifa = &ia->ia_ifa;
/*
* Prepare info data for the host route.
* This code mimics one from ifa_maintain_loopback_route().
*/
bzero(&info, sizeof(struct rt_addrinfo));
info.rti_flags = ifa->ifa_flags | RTF_HOST | RTF_STATIC | RTF_PINNED;
info.rti_info[RTAX_DST] = ifa->ifa_addr;
info.rti_info[RTAX_GATEWAY] = (struct sockaddr *)&gateway;
link_init_sdl(ifa->ifa_ifp, (struct sockaddr *)&gateway, ifa->ifa_ifp->if_type);
if (cmd != RTM_DELETE)
info.rti_ifp = V_loif;
fibnum = ia62ifa(ia)->ifa_ifp->if_fib;
if (cmd == RTM_ADD) {
rt_addrmsg(cmd, &ia->ia_ifa, fibnum);
rt_routemsg_info(cmd, &info, fibnum);
} else if (cmd == RTM_DELETE) {
rt_routemsg_info(cmd, &info, fibnum);
rt_addrmsg(cmd, &ia->ia_ifa, fibnum);
}
}
int
in6_mask2len(struct in6_addr *mask, u_char *lim0)
{
int x = 0, y;
u_char *lim = lim0, *p;
/* ignore the scope_id part */
if (lim0 == NULL || lim0 - (u_char *)mask > sizeof(*mask))
lim = (u_char *)mask + sizeof(*mask);
for (p = (u_char *)mask; p < lim; x++, p++) {
if (*p != 0xff)
break;
}
y = 0;
if (p < lim) {
for (y = 0; y < 8; y++) {
if ((*p & (0x80 >> y)) == 0)
break;
}
}
/*
* when the limit pointer is given, do a stricter check on the
* remaining bits.
*/
if (p < lim) {
if (y != 0 && (*p & (0x00ff >> y)) != 0)
return (-1);
for (p = p + 1; p < lim; p++)
if (*p != 0)
return (-1);
}
return x * 8 + y;
}
#ifdef COMPAT_FREEBSD32
struct in6_ndifreq32 {
char ifname[IFNAMSIZ];
uint32_t ifindex;
};
#define SIOCGDEFIFACE32_IN6 _IOWR('i', 86, struct in6_ndifreq32)
#endif
int
in6_control(struct socket *so, u_long cmd, caddr_t data,
struct ifnet *ifp, struct thread *td)
{
struct in6_ifreq *ifr = (struct in6_ifreq *)data;
struct in6_ifaddr *ia = NULL;
struct in6_aliasreq *ifra = (struct in6_aliasreq *)data;
struct sockaddr_in6 *sa6;
int carp_attached = 0;
int error;
u_long ocmd = cmd;
/*
* Compat to make pre-10.x ifconfig(8) operable.
*/
if (cmd == OSIOCAIFADDR_IN6)
cmd = SIOCAIFADDR_IN6;
switch (cmd) {
case SIOCGETSGCNT_IN6:
case SIOCGETMIFCNT_IN6:
/*
* XXX mrt_ioctl has a 3rd, unused, FIB argument in route.c.
* We cannot see how that would be needed, so do not adjust the
* KPI blindly; more likely should clean up the IPv4 variant.
*/
return (mrt6_ioctl ? mrt6_ioctl(cmd, data) : EOPNOTSUPP);
}
switch (cmd) {
case SIOCAADDRCTL_POLICY:
case SIOCDADDRCTL_POLICY:
if (td != NULL) {
error = priv_check(td, PRIV_NETINET_ADDRCTRL6);
if (error)
return (error);
}
return (in6_src_ioctl(cmd, data));
}
if (ifp == NULL)
return (EOPNOTSUPP);
switch (cmd) {
case SIOCSNDFLUSH_IN6:
case SIOCSPFXFLUSH_IN6:
case SIOCSRTRFLUSH_IN6:
case SIOCSDEFIFACE_IN6:
case SIOCSIFINFO_FLAGS:
case SIOCSIFINFO_IN6:
if (td != NULL) {
error = priv_check(td, PRIV_NETINET_ND6);
if (error)
return (error);
}
/* FALLTHROUGH */
case OSIOCGIFINFO_IN6:
case SIOCGIFINFO_IN6:
case SIOCGNBRINFO_IN6:
case SIOCGDEFIFACE_IN6:
return (nd6_ioctl(cmd, data, ifp));
#ifdef COMPAT_FREEBSD32
case SIOCGDEFIFACE32_IN6:
{
struct in6_ndifreq ndif;
struct in6_ndifreq32 *ndif32;
error = nd6_ioctl(SIOCGDEFIFACE_IN6, (caddr_t)&ndif,
ifp);
if (error)
return (error);
ndif32 = (struct in6_ndifreq32 *)data;
ndif32->ifindex = ndif.ifindex;
return (0);
}
#endif
}
switch (cmd) {
case SIOCSIFPREFIX_IN6:
case SIOCDIFPREFIX_IN6:
case SIOCAIFPREFIX_IN6:
case SIOCCIFPREFIX_IN6:
case SIOCSGIFPREFIX_IN6:
case SIOCGIFPREFIX_IN6:
log(LOG_NOTICE,
"prefix ioctls are now invalidated. "
"please use ifconfig.\n");
return (EOPNOTSUPP);
}
switch (cmd) {
case SIOCSSCOPE6:
if (td != NULL) {
error = priv_check(td, PRIV_NETINET_SCOPE6);
if (error)
return (error);
}
/* FALLTHROUGH */
case SIOCGSCOPE6:
case SIOCGSCOPE6DEF:
return (scope6_ioctl(cmd, data, ifp));
}
/*
* Find address for this interface, if it exists.
*
* In netinet code, we have checked ifra_addr in SIOCSIF*ADDR operation
* only, and used the first interface address as the target of other
* operations (without checking ifra_addr). This was because netinet
* code/API assumed at most 1 interface address per interface.
* Since IPv6 allows a node to assign multiple addresses
* on a single interface, we almost always look and check the
* presence of ifra_addr, and reject invalid ones here.
* It also decreases duplicated code among SIOC*_IN6 operations.
*/
switch (cmd) {
case SIOCAIFADDR_IN6:
case SIOCSIFPHYADDR_IN6:
sa6 = &ifra->ifra_addr;
break;
case SIOCSIFADDR_IN6:
case SIOCGIFADDR_IN6:
case SIOCSIFDSTADDR_IN6:
case SIOCSIFNETMASK_IN6:
case SIOCGIFDSTADDR_IN6:
case SIOCGIFNETMASK_IN6:
case SIOCDIFADDR_IN6:
case SIOCGIFPSRCADDR_IN6:
case SIOCGIFPDSTADDR_IN6:
case SIOCGIFAFLAG_IN6:
case SIOCSNDFLUSH_IN6:
case SIOCSPFXFLUSH_IN6:
case SIOCSRTRFLUSH_IN6:
case SIOCGIFALIFETIME_IN6:
case SIOCGIFSTAT_IN6:
case SIOCGIFSTAT_ICMP6:
sa6 = &ifr->ifr_addr;
break;
case SIOCSIFADDR:
case SIOCSIFBRDADDR:
case SIOCSIFDSTADDR:
case SIOCSIFNETMASK:
/*
* Although we should pass any non-INET6 ioctl requests
* down to driver, we filter some legacy INET requests.
* Drivers trust SIOCSIFADDR et al to come from an already
* privileged layer, and do not perform any credentials
* checks or input validation.
*/
return (EINVAL);
default:
sa6 = NULL;
break;
}
if (sa6 && sa6->sin6_family == AF_INET6) {
if (sa6->sin6_scope_id != 0)
error = sa6_embedscope(sa6, 0);
else
error = in6_setscope(&sa6->sin6_addr, ifp, NULL);
if (error != 0)
return (error);
if (td != NULL && (error = prison_check_ip6(td->td_ucred,
&sa6->sin6_addr)) != 0)
return (error);
ia = in6ifa_ifpwithaddr(ifp, &sa6->sin6_addr);
} else
ia = NULL;
switch (cmd) {
case SIOCSIFADDR_IN6:
case SIOCSIFDSTADDR_IN6:
case SIOCSIFNETMASK_IN6:
/*
* Since IPv6 allows a node to assign multiple addresses
* on a single interface, SIOCSIFxxx ioctls are deprecated.
*/
/* we decided to obsolete this command (20000704) */
error = EINVAL;
goto out;
case SIOCDIFADDR_IN6:
/*
* for IPv4, we look for existing in_ifaddr here to allow
* "ifconfig if0 delete" to remove the first IPv4 address on
* the interface. For IPv6, as the spec allows multiple
* interface address from the day one, we consider "remove the
* first one" semantics to be not preferable.
*/
if (ia == NULL) {
error = EADDRNOTAVAIL;
goto out;
}
/* FALLTHROUGH */
case SIOCAIFADDR_IN6:
/*
* We always require users to specify a valid IPv6 address for
* the corresponding operation.
*/
if (ifra->ifra_addr.sin6_family != AF_INET6 ||
ifra->ifra_addr.sin6_len != sizeof(struct sockaddr_in6)) {
error = EAFNOSUPPORT;
goto out;
}
if (td != NULL) {
error = priv_check(td, (cmd == SIOCDIFADDR_IN6) ?
PRIV_NET_DELIFADDR : PRIV_NET_ADDIFADDR);
if (error)
goto out;
}
/* FALLTHROUGH */
case SIOCGIFSTAT_IN6:
case SIOCGIFSTAT_ICMP6:
if (ifp->if_afdata[AF_INET6] == NULL) {
error = EPFNOSUPPORT;
goto out;
}
break;
case SIOCGIFADDR_IN6:
/* This interface is basically deprecated. use SIOCGIFCONF. */
/* FALLTHROUGH */
case SIOCGIFAFLAG_IN6:
case SIOCGIFNETMASK_IN6:
case SIOCGIFDSTADDR_IN6:
case SIOCGIFALIFETIME_IN6:
/* must think again about its semantics */
if (ia == NULL) {
error = EADDRNOTAVAIL;
goto out;
}
break;
}
switch (cmd) {
case SIOCGIFADDR_IN6:
ifr->ifr_addr = ia->ia_addr;
if ((error = sa6_recoverscope(&ifr->ifr_addr)) != 0)
goto out;
break;
case SIOCGIFDSTADDR_IN6:
if ((ifp->if_flags & IFF_POINTOPOINT) == 0) {
error = EINVAL;
goto out;
}
ifr->ifr_dstaddr = ia->ia_dstaddr;
if ((error = sa6_recoverscope(&ifr->ifr_dstaddr)) != 0)
goto out;
break;
case SIOCGIFNETMASK_IN6:
ifr->ifr_addr = ia->ia_prefixmask;
break;
case SIOCGIFAFLAG_IN6:
ifr->ifr_ifru.ifru_flags6 = ia->ia6_flags;
break;
case SIOCGIFSTAT_IN6:
COUNTER_ARRAY_COPY(((struct in6_ifextra *)
ifp->if_afdata[AF_INET6])->in6_ifstat,
&ifr->ifr_ifru.ifru_stat,
sizeof(struct in6_ifstat) / sizeof(uint64_t));
break;
case SIOCGIFSTAT_ICMP6:
COUNTER_ARRAY_COPY(((struct in6_ifextra *)
ifp->if_afdata[AF_INET6])->icmp6_ifstat,
&ifr->ifr_ifru.ifru_icmp6stat,
sizeof(struct icmp6_ifstat) / sizeof(uint64_t));
break;
case SIOCGIFALIFETIME_IN6:
ifr->ifr_ifru.ifru_lifetime = ia->ia6_lifetime;
if (ia->ia6_lifetime.ia6t_vltime != ND6_INFINITE_LIFETIME) {
time_t maxexpire;
struct in6_addrlifetime *retlt =
&ifr->ifr_ifru.ifru_lifetime;
/*
* XXX: adjust expiration time assuming time_t is
* signed.
*/
maxexpire = (-1) &
~((time_t)1 << ((sizeof(maxexpire) * 8) - 1));
if (ia->ia6_lifetime.ia6t_vltime <
maxexpire - ia->ia6_updatetime) {
retlt->ia6t_expire = ia->ia6_updatetime +
ia->ia6_lifetime.ia6t_vltime;
} else
retlt->ia6t_expire = maxexpire;
}
if (ia->ia6_lifetime.ia6t_pltime != ND6_INFINITE_LIFETIME) {
time_t maxexpire;
struct in6_addrlifetime *retlt =
&ifr->ifr_ifru.ifru_lifetime;
/*
* XXX: adjust expiration time assuming time_t is
* signed.
*/
maxexpire = (-1) &
~((time_t)1 << ((sizeof(maxexpire) * 8) - 1));
if (ia->ia6_lifetime.ia6t_pltime <
maxexpire - ia->ia6_updatetime) {
retlt->ia6t_preferred = ia->ia6_updatetime +
ia->ia6_lifetime.ia6t_pltime;
} else
retlt->ia6t_preferred = maxexpire;
}
break;
case SIOCAIFADDR_IN6:
{
struct nd_prefixctl pr0;
struct nd_prefix *pr;
/*
* first, make or update the interface address structure,
* and link it to the list.
*/
if ((error = in6_update_ifa(ifp, ifra, ia, 0)) != 0)
goto out;
if (ia != NULL) {
if (ia->ia_ifa.ifa_carp)
(*carp_detach_p)(&ia->ia_ifa, true);
ifa_free(&ia->ia_ifa);
}
if ((ia = in6ifa_ifpwithaddr(ifp, &ifra->ifra_addr.sin6_addr))
== NULL) {
/*
* this can happen when the user specify the 0 valid
* lifetime.
*/
break;
}
if (cmd == ocmd && ifra->ifra_vhid > 0) {
if (carp_attach_p != NULL)
error = (*carp_attach_p)(&ia->ia_ifa,
ifra->ifra_vhid);
else
error = EPROTONOSUPPORT;
if (error)
goto out;
else
carp_attached = 1;
}
/*
* then, make the prefix on-link on the interface.
* XXX: we'd rather create the prefix before the address, but
* we need at least one address to install the corresponding
* interface route, so we configure the address first.
*/
/*
* convert mask to prefix length (prefixmask has already
* been validated in in6_update_ifa().
*/
bzero(&pr0, sizeof(pr0));
pr0.ndpr_ifp = ifp;
pr0.ndpr_plen = in6_mask2len(&ifra->ifra_prefixmask.sin6_addr,
NULL);
if (pr0.ndpr_plen == 128) {
/* we don't need to install a host route. */
goto aifaddr_out;
}
pr0.ndpr_prefix = ifra->ifra_addr;
/* apply the mask for safety. */
IN6_MASK_ADDR(&pr0.ndpr_prefix.sin6_addr,
&ifra->ifra_prefixmask.sin6_addr);
/*
* XXX: since we don't have an API to set prefix (not address)
* lifetimes, we just use the same lifetimes as addresses.
* The (temporarily) installed lifetimes can be overridden by
* later advertised RAs (when accept_rtadv is non 0), which is
* an intended behavior.
*/
pr0.ndpr_raf_onlink = 1; /* should be configurable? */
pr0.ndpr_raf_auto =
((ifra->ifra_flags & IN6_IFF_AUTOCONF) != 0);
pr0.ndpr_vltime = ifra->ifra_lifetime.ia6t_vltime;
pr0.ndpr_pltime = ifra->ifra_lifetime.ia6t_pltime;
/* add the prefix if not yet. */
if ((pr = nd6_prefix_lookup(&pr0)) == NULL) {
/*
* nd6_prelist_add will install the corresponding
* interface route.
*/
if ((error = nd6_prelist_add(&pr0, NULL, &pr)) != 0) {
if (carp_attached)
(*carp_detach_p)(&ia->ia_ifa, false);
goto out;
}
}
/* relate the address to the prefix */
if (ia->ia6_ndpr == NULL) {
ia->ia6_ndpr = pr;
pr->ndpr_addrcnt++;
/*
* If this is the first autoconf address from the
* prefix, create a temporary address as well
* (when required).
*/
if ((ia->ia6_flags & IN6_IFF_AUTOCONF) &&
V_ip6_use_tempaddr && pr->ndpr_addrcnt == 1) {
int e;
if ((e = in6_tmpifadd(ia, 1, 0)) != 0) {
log(LOG_NOTICE, "in6_control: failed "
"to create a temporary address, "
"errno=%d\n", e);
}
}
}
nd6_prefix_rele(pr);
/*
* this might affect the status of autoconfigured addresses,
* that is, this address might make other addresses detached.
*/
pfxlist_onlink_check();
aifaddr_out:
/*
* Try to clear the flag when a new IPv6 address is added
* onto an IFDISABLED interface and it succeeds.
*/
if (ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) {
struct in6_ndireq nd;
memset(&nd, 0, sizeof(nd));
nd.ndi.flags = ND_IFINFO(ifp)->flags;
nd.ndi.flags &= ~ND6_IFF_IFDISABLED;
if (nd6_ioctl(SIOCSIFINFO_FLAGS, (caddr_t)&nd, ifp) < 0)
log(LOG_NOTICE, "SIOCAIFADDR_IN6: "
"SIOCSIFINFO_FLAGS for -ifdisabled "
"failed.");
/*
* Ignore failure of clearing the flag intentionally.
* The failure means address duplication was detected.
*/
}
break;
}
case SIOCDIFADDR_IN6:
in6_purgeifaddr(ia);
EVENTHANDLER_INVOKE(ifaddr_event_ext, ifp, &ia->ia_ifa,
IFADDR_EVENT_DEL);
break;
default:
if (ifp->if_ioctl == NULL) {
error = EOPNOTSUPP;
goto out;
}
error = (*ifp->if_ioctl)(ifp, cmd, data);
goto out;
}
error = 0;
out:
if (ia != NULL)
ifa_free(&ia->ia_ifa);
return (error);
}
static struct in6_multi_mship *
in6_joingroup_legacy(struct ifnet *ifp, const struct in6_addr *mcaddr,
int *errorp, int delay)
{
struct in6_multi_mship *imm;
int error;
imm = malloc(sizeof(*imm), M_IP6MADDR, M_NOWAIT);
if (imm == NULL) {
*errorp = ENOBUFS;
return (NULL);
}
delay = (delay * PR_FASTHZ) / hz;
error = in6_joingroup(ifp, mcaddr, NULL, &imm->i6mm_maddr, delay);
if (error) {
*errorp = error;
free(imm, M_IP6MADDR);
return (NULL);
}
return (imm);
}
/*
* Join necessary multicast groups. Factored out from in6_update_ifa().
* This entire work should only be done once, for the default FIB.
*/
static int
in6_update_ifa_join_mc(struct ifnet *ifp, struct in6_aliasreq *ifra,
struct in6_ifaddr *ia, int flags, struct in6_multi **in6m_sol)
{
char ip6buf[INET6_ADDRSTRLEN];
struct in6_addr mltaddr;
struct in6_multi_mship *imm;
int delay, error;
KASSERT(in6m_sol != NULL, ("%s: in6m_sol is NULL", __func__));
/* Join solicited multicast addr for new host id. */
bzero(&mltaddr, sizeof(struct in6_addr));
mltaddr.s6_addr32[0] = IPV6_ADDR_INT32_MLL;
mltaddr.s6_addr32[2] = htonl(1);
mltaddr.s6_addr32[3] = ifra->ifra_addr.sin6_addr.s6_addr32[3];
mltaddr.s6_addr8[12] = 0xff;
if ((error = in6_setscope(&mltaddr, ifp, NULL)) != 0) {
/* XXX: should not happen */
log(LOG_ERR, "%s: in6_setscope failed\n", __func__);
goto cleanup;
}
delay = error = 0;
if ((flags & IN6_IFAUPDATE_DADDELAY)) {
/*
* We need a random delay for DAD on the address being
* configured. It also means delaying transmission of the
* corresponding MLD report to avoid report collision.
* [RFC 4861, Section 6.3.7]
*/
delay = arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz);
}
imm = in6_joingroup_legacy(ifp, &mltaddr, &error, delay);
if (imm == NULL) {
nd6log((LOG_WARNING, "%s: in6_joingroup failed for %s on %s "
"(errno=%d)\n", __func__, ip6_sprintf(ip6buf, &mltaddr),
if_name(ifp), error));
goto cleanup;
}
LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
*in6m_sol = imm->i6mm_maddr;
/*
* Join link-local all-nodes address.
*/
mltaddr = in6addr_linklocal_allnodes;
if ((error = in6_setscope(&mltaddr, ifp, NULL)) != 0)
goto cleanup; /* XXX: should not fail */
imm = in6_joingroup_legacy(ifp, &mltaddr, &error, 0);
if (imm == NULL) {
nd6log((LOG_WARNING, "%s: in6_joingroup failed for %s on %s "
"(errno=%d)\n", __func__, ip6_sprintf(ip6buf, &mltaddr),
if_name(ifp), error));
goto cleanup;
}
LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
/*
* Join node information group address.
*/
delay = 0;
if ((flags & IN6_IFAUPDATE_DADDELAY)) {
/*
* The spec does not say anything about delay for this group,
* but the same logic should apply.
*/
delay = arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz);
}
if (in6_nigroup(ifp, NULL, -1, &mltaddr) == 0) {
/* XXX jinmei */
imm = in6_joingroup_legacy(ifp, &mltaddr, &error, delay);
if (imm == NULL)
nd6log((LOG_WARNING,
"%s: in6_joingroup failed for %s on %s "
"(errno=%d)\n", __func__, ip6_sprintf(ip6buf,
&mltaddr), if_name(ifp), error));
/* XXX not very fatal, go on... */
else
LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
}
if (V_icmp6_nodeinfo_oldmcprefix &&
in6_nigroup_oldmcprefix(ifp, NULL, -1, &mltaddr) == 0) {
imm = in6_joingroup_legacy(ifp, &mltaddr, &error, delay);
if (imm == NULL)
nd6log((LOG_WARNING,
"%s: in6_joingroup failed for %s on %s "
"(errno=%d)\n", __func__, ip6_sprintf(ip6buf,
&mltaddr), if_name(ifp), error));
/* XXX not very fatal, go on... */
else
LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
}
/*
* Join interface-local all-nodes address.
* (ff01::1%ifN, and ff01::%ifN/32)
*/
mltaddr = in6addr_nodelocal_allnodes;
if ((error = in6_setscope(&mltaddr, ifp, NULL)) != 0)
goto cleanup; /* XXX: should not fail */
imm = in6_joingroup_legacy(ifp, &mltaddr, &error, 0);
if (imm == NULL) {
nd6log((LOG_WARNING, "%s: in6_joingroup failed for %s on %s "
"(errno=%d)\n", __func__, ip6_sprintf(ip6buf,
&mltaddr), if_name(ifp), error));
goto cleanup;
}
LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
cleanup:
return (error);
}
/*
* Update parameters of an IPv6 interface address.
* If necessary, a new entry is created and linked into address chains.
* This function is separated from in6_control().
*/
int
in6_update_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra,
struct in6_ifaddr *ia, int flags)
{
int error, hostIsNew = 0;
if ((error = in6_validate_ifra(ifp, ifra, ia, flags)) != 0)
return (error);
if (ia == NULL) {
hostIsNew = 1;
if ((ia = in6_alloc_ifa(ifp, ifra, flags)) == NULL)
return (ENOBUFS);
}
error = in6_update_ifa_internal(ifp, ifra, ia, hostIsNew, flags);
if (error != 0) {
if (hostIsNew != 0) {
in6_unlink_ifa(ia, ifp);
ifa_free(&ia->ia_ifa);
}
return (error);
}
if (hostIsNew)
error = in6_broadcast_ifa(ifp, ifra, ia, flags);
return (error);
}
/*
* Fill in basic IPv6 address request info.
*/
void
in6_prepare_ifra(struct in6_aliasreq *ifra, const struct in6_addr *addr,
const struct in6_addr *mask)
{
memset(ifra, 0, sizeof(struct in6_aliasreq));
ifra->ifra_addr.sin6_family = AF_INET6;
ifra->ifra_addr.sin6_len = sizeof(struct sockaddr_in6);
if (addr != NULL)
ifra->ifra_addr.sin6_addr = *addr;
ifra->ifra_prefixmask.sin6_family = AF_INET6;
ifra->ifra_prefixmask.sin6_len = sizeof(struct sockaddr_in6);
if (mask != NULL)
ifra->ifra_prefixmask.sin6_addr = *mask;
}
static int
in6_validate_ifra(struct ifnet *ifp, struct in6_aliasreq *ifra,
struct in6_ifaddr *ia, int flags)
{
int plen = -1;
struct sockaddr_in6 dst6;
struct in6_addrlifetime *lt;
char ip6buf[INET6_ADDRSTRLEN];
/* Validate parameters */
if (ifp == NULL || ifra == NULL) /* this maybe redundant */
return (EINVAL);
/*
* The destination address for a p2p link must have a family
* of AF_UNSPEC or AF_INET6.
*/
if ((ifp->if_flags & IFF_POINTOPOINT) != 0 &&
ifra->ifra_dstaddr.sin6_family != AF_INET6 &&
ifra->ifra_dstaddr.sin6_family != AF_UNSPEC)
return (EAFNOSUPPORT);
/*
* Validate address
*/
if (ifra->ifra_addr.sin6_len != sizeof(struct sockaddr_in6) ||
ifra->ifra_addr.sin6_family != AF_INET6)
return (EINVAL);
/*
* validate ifra_prefixmask. don't check sin6_family, netmask
* does not carry fields other than sin6_len.
*/
if (ifra->ifra_prefixmask.sin6_len > sizeof(struct sockaddr_in6))
return (EINVAL);
/*
* Because the IPv6 address architecture is classless, we require
* users to specify a (non 0) prefix length (mask) for a new address.
* We also require the prefix (when specified) mask is valid, and thus
* reject a non-consecutive mask.
*/
if (ia == NULL && ifra->ifra_prefixmask.sin6_len == 0)
return (EINVAL);
if (ifra->ifra_prefixmask.sin6_len != 0) {
plen = in6_mask2len(&ifra->ifra_prefixmask.sin6_addr,
(u_char *)&ifra->ifra_prefixmask +
ifra->ifra_prefixmask.sin6_len);
if (plen <= 0)
return (EINVAL);
} else {
/*
* In this case, ia must not be NULL. We just use its prefix
* length.
*/
plen = in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL);
}
/*
* If the destination address on a p2p interface is specified,
* and the address is a scoped one, validate/set the scope
* zone identifier.
*/
dst6 = ifra->ifra_dstaddr;
if ((ifp->if_flags & (IFF_POINTOPOINT|IFF_LOOPBACK)) != 0 &&
(dst6.sin6_family == AF_INET6)) {
struct in6_addr in6_tmp;
u_int32_t zoneid;
in6_tmp = dst6.sin6_addr;
if (in6_setscope(&in6_tmp, ifp, &zoneid))
return (EINVAL); /* XXX: should be impossible */
if (dst6.sin6_scope_id != 0) {
if (dst6.sin6_scope_id != zoneid)
return (EINVAL);
} else /* user omit to specify the ID. */
dst6.sin6_scope_id = zoneid;
/* convert into the internal form */
if (sa6_embedscope(&dst6, 0))
return (EINVAL); /* XXX: should be impossible */
}
/* Modify original ifra_dstaddr to reflect changes */
ifra->ifra_dstaddr = dst6;
/*
* The destination address can be specified only for a p2p or a
* loopback interface. If specified, the corresponding prefix length
* must be 128.
*/
if (ifra->ifra_dstaddr.sin6_family == AF_INET6) {
if ((ifp->if_flags & (IFF_POINTOPOINT|IFF_LOOPBACK)) == 0) {
/* XXX: noisy message */
nd6log((LOG_INFO, "in6_update_ifa: a destination can "
"be specified for a p2p or a loopback IF only\n"));
return (EINVAL);
}
if (plen != 128) {
nd6log((LOG_INFO, "in6_update_ifa: prefixlen should "
"be 128 when dstaddr is specified\n"));
return (EINVAL);
}
}
/* lifetime consistency check */
lt = &ifra->ifra_lifetime;
if (lt->ia6t_pltime > lt->ia6t_vltime)
return (EINVAL);
if (lt->ia6t_vltime == 0) {
/*
* the following log might be noisy, but this is a typical
* configuration mistake or a tool's bug.
*/
nd6log((LOG_INFO,
"in6_update_ifa: valid lifetime is 0 for %s\n",
ip6_sprintf(ip6buf, &ifra->ifra_addr.sin6_addr)));
if (ia == NULL)
return (0); /* there's nothing to do */
}
/* Check prefix mask */
if (ia != NULL && ifra->ifra_prefixmask.sin6_len != 0) {
/*
* We prohibit changing the prefix length of an existing
* address, because
* + such an operation should be rare in IPv6, and
* + the operation would confuse prefix management.
*/
if (ia->ia_prefixmask.sin6_len != 0 &&
in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL) != plen) {
nd6log((LOG_INFO, "in6_validate_ifa: the prefix length "
"of an existing %s address should not be changed\n",
ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr)));
return (EINVAL);
}
}
return (0);
}
/*
* Allocate a new ifaddr and link it into chains.
*/
static struct in6_ifaddr *
in6_alloc_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra, int flags)
{
struct in6_ifaddr *ia;
/*
* When in6_alloc_ifa() is called in a process of a received
* RA, it is called under an interrupt context. So, we should
* call malloc with M_NOWAIT.
*/
ia = (struct in6_ifaddr *)ifa_alloc(sizeof(*ia), M_NOWAIT);
if (ia == NULL)
return (NULL);
LIST_INIT(&ia->ia6_memberships);
/* Initialize the address and masks, and put time stamp */
ia->ia_ifa.ifa_addr = (struct sockaddr *)&ia->ia_addr;
ia->ia_addr.sin6_family = AF_INET6;
ia->ia_addr.sin6_len = sizeof(ia->ia_addr);
/* XXX: Can we assign ,sin6_addr and skip the rest? */
ia->ia_addr = ifra->ifra_addr;
ia->ia6_createtime = time_uptime;
if ((ifp->if_flags & (IFF_POINTOPOINT | IFF_LOOPBACK)) != 0) {
/*
* Some functions expect that ifa_dstaddr is not
* NULL for p2p interfaces.
*/
ia->ia_ifa.ifa_dstaddr =
(struct sockaddr *)&ia->ia_dstaddr;
} else {
ia->ia_ifa.ifa_dstaddr = NULL;
}
/* set prefix mask if any */
ia->ia_ifa.ifa_netmask = (struct sockaddr *)&ia->ia_prefixmask;
if (ifra->ifra_prefixmask.sin6_len != 0) {
ia->ia_prefixmask.sin6_family = AF_INET6;
ia->ia_prefixmask.sin6_len = ifra->ifra_prefixmask.sin6_len;
ia->ia_prefixmask.sin6_addr = ifra->ifra_prefixmask.sin6_addr;
}
ia->ia_ifp = ifp;
ifa_ref(&ia->ia_ifa); /* if_addrhead */
IF_ADDR_WLOCK(ifp);
CK_STAILQ_INSERT_TAIL(&ifp->if_addrhead, &ia->ia_ifa, ifa_link);
IF_ADDR_WUNLOCK(ifp);
ifa_ref(&ia->ia_ifa); /* in6_ifaddrhead */
IN6_IFADDR_WLOCK();
CK_STAILQ_INSERT_TAIL(&V_in6_ifaddrhead, ia, ia_link);
CK_LIST_INSERT_HEAD(IN6ADDR_HASH(&ia->ia_addr.sin6_addr), ia, ia6_hash);
IN6_IFADDR_WUNLOCK();
return (ia);
}
/*
* Update/configure interface address parameters:
*
* 1) Update lifetime
* 2) Update interface metric ad flags
* 3) Notify other subsystems
*/
static int
in6_update_ifa_internal(struct ifnet *ifp, struct in6_aliasreq *ifra,
struct in6_ifaddr *ia, int hostIsNew, int flags)
{
int error;
/* update timestamp */
ia->ia6_updatetime = time_uptime;
/*
* Set lifetimes. We do not refer to ia6t_expire and ia6t_preferred
* to see if the address is deprecated or invalidated, but initialize
* these members for applications.
*/
ia->ia6_lifetime = ifra->ifra_lifetime;
if (ia->ia6_lifetime.ia6t_vltime != ND6_INFINITE_LIFETIME) {
ia->ia6_lifetime.ia6t_expire =
time_uptime + ia->ia6_lifetime.ia6t_vltime;
} else
ia->ia6_lifetime.ia6t_expire = 0;
if (ia->ia6_lifetime.ia6t_pltime != ND6_INFINITE_LIFETIME) {
ia->ia6_lifetime.ia6t_preferred =
time_uptime + ia->ia6_lifetime.ia6t_pltime;
} else
ia->ia6_lifetime.ia6t_preferred = 0;
/*
* backward compatibility - if IN6_IFF_DEPRECATED is set from the
* userland, make it deprecated.
*/
if ((ifra->ifra_flags & IN6_IFF_DEPRECATED) != 0) {
ia->ia6_lifetime.ia6t_pltime = 0;
ia->ia6_lifetime.ia6t_preferred = time_uptime;
}
/*
* configure address flags.
*/
ia->ia6_flags = ifra->ifra_flags;
/*
* Make the address tentative before joining multicast addresses,
* so that corresponding MLD responses would not have a tentative
* source address.
*/
ia->ia6_flags &= ~IN6_IFF_DUPLICATED; /* safety */
/*
* DAD should be performed for an new address or addresses on
* an interface with ND6_IFF_IFDISABLED.
*/
if (in6if_do_dad(ifp) &&
(hostIsNew || (ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED)))
ia->ia6_flags |= IN6_IFF_TENTATIVE;
/* notify other subsystems */
error = in6_notify_ifa(ifp, ia, ifra, hostIsNew);
return (error);
}
/*
* Do link-level ifa job:
* 1) Add lle entry for added address
* 2) Notifies routing socket users about new address
* 3) join appropriate multicast group
* 4) start DAD if enabled
*/
static int
in6_broadcast_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra,
struct in6_ifaddr *ia, int flags)
{
struct in6_multi *in6m_sol;
int error = 0;
/* Add local address to lltable, if necessary (ex. on p2p link). */
if ((error = nd6_add_ifa_lle(ia)) != 0) {
in6_purgeaddr(&ia->ia_ifa);
ifa_free(&ia->ia_ifa);
return (error);
}
/* Join necessary multicast groups. */
in6m_sol = NULL;
if ((ifp->if_flags & IFF_MULTICAST) != 0) {
error = in6_update_ifa_join_mc(ifp, ifra, ia, flags, &in6m_sol);
if (error != 0) {
in6_purgeaddr(&ia->ia_ifa);
ifa_free(&ia->ia_ifa);
return (error);
}
}
/* Perform DAD, if the address is TENTATIVE. */
if ((ia->ia6_flags & IN6_IFF_TENTATIVE)) {
int delay, mindelay, maxdelay;
delay = 0;
if ((flags & IN6_IFAUPDATE_DADDELAY)) {
/*
* We need to impose a delay before sending an NS
* for DAD. Check if we also needed a delay for the
* corresponding MLD message. If we did, the delay
* should be larger than the MLD delay (this could be
* relaxed a bit, but this simple logic is at least
* safe).
* XXX: Break data hiding guidelines and look at
* state for the solicited multicast group.
*/
mindelay = 0;
if (in6m_sol != NULL &&
in6m_sol->in6m_state == MLD_REPORTING_MEMBER) {
mindelay = in6m_sol->in6m_timer;
}
maxdelay = MAX_RTR_SOLICITATION_DELAY * hz;
if (maxdelay - mindelay == 0)
delay = 0;
else {
delay =
(arc4random() % (maxdelay - mindelay)) +
mindelay;
}
}
nd6_dad_start((struct ifaddr *)ia, delay);
}
in6_newaddrmsg(ia, RTM_ADD);
ifa_free(&ia->ia_ifa);
return (error);
}
/*
* Adds or deletes interface route for p2p ifa.
* Returns 0 on success or errno.
*/
static int
in6_handle_dstaddr_rtrequest(int cmd, struct in6_ifaddr *ia)
{
struct epoch_tracker et;
struct ifaddr *ifa = &ia->ia_ifa;
int error;
/* Prepare gateway */
struct sockaddr_dl_short sdl = {
.sdl_family = AF_LINK,
.sdl_len = sizeof(struct sockaddr_dl_short),
.sdl_type = ifa->ifa_ifp->if_type,
.sdl_index = ifa->ifa_ifp->if_index,
};
struct sockaddr_in6 dst = {
.sin6_family = AF_INET6,
.sin6_len = sizeof(struct sockaddr_in6),
.sin6_addr = ia->ia_dstaddr.sin6_addr,
};
struct rt_addrinfo info = {
.rti_ifa = ifa,
.rti_flags = RTF_PINNED | RTF_HOST,
.rti_info = {
[RTAX_DST] = (struct sockaddr *)&dst,
[RTAX_GATEWAY] = (struct sockaddr *)&sdl,
},
};
/* Don't set additional per-gw filters on removal */
NET_EPOCH_ENTER(et);
error = rib_handle_ifaddr_info(ifa->ifa_ifp->if_fib, cmd, &info);
NET_EPOCH_EXIT(et);
return (error);
}
static bool
ifa_is_p2p(struct in6_ifaddr *ia)
{
int plen;
plen = in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL); /* XXX */
if ((plen == 128) && (ia->ia_dstaddr.sin6_family == AF_INET6) &&
!IN6_ARE_ADDR_EQUAL(&ia->ia_addr.sin6_addr, &ia->ia_dstaddr.sin6_addr))
return (true);
return (false);
}
void
in6_purgeaddr(struct ifaddr *ifa)
{
struct ifnet *ifp = ifa->ifa_ifp;
struct in6_ifaddr *ia = (struct in6_ifaddr *) ifa;
struct in6_multi_mship *imm;
int error;
if (ifa->ifa_carp)
(*carp_detach_p)(ifa, false);
/*
* Remove the loopback route to the interface address.
* The check for the current setting of "nd6_useloopback"
* is not needed.
*/
if (ia->ia_flags & IFA_RTSELF) {
error = ifa_del_loopback_route((struct ifaddr *)ia,
(struct sockaddr *)&ia->ia_addr);
if (error == 0)
ia->ia_flags &= ~IFA_RTSELF;
}
/* stop DAD processing */
nd6_dad_stop(ifa);
/* Leave multicast groups. */
while ((imm = LIST_FIRST(&ia->ia6_memberships)) != NULL) {
LIST_REMOVE(imm, i6mm_chain);
if (imm->i6mm_maddr != NULL)
in6_leavegroup(imm->i6mm_maddr, NULL);
free(imm, M_IP6MADDR);
}
/* Check if we need to remove p2p route */
if ((ia->ia_flags & IFA_ROUTE) && ifa_is_p2p(ia)) {
error = in6_handle_dstaddr_rtrequest(RTM_DELETE, ia);
if (error != 0)
log(LOG_INFO, "%s: err=%d, destination address delete "
"failed\n", __func__, error);
ia->ia_flags &= ~IFA_ROUTE;
}
in6_newaddrmsg(ia, RTM_DELETE);
in6_unlink_ifa(ia, ifp);
}
/*
* Removes @ia from the corresponding interfaces and unlinks corresponding
* prefix if no addresses are using it anymore.
*/
void
in6_purgeifaddr(struct in6_ifaddr *ia)
{
struct nd_prefix *pr;
/*
* If the address being deleted is the only one that owns
* the corresponding prefix, expire the prefix as well.
* XXX: theoretically, we don't have to worry about such
* relationship, since we separate the address management
* and the prefix management. We do this, however, to provide
* as much backward compatibility as possible in terms of
* the ioctl operation.
* Note that in6_purgeaddr() will decrement ndpr_addrcnt.
*/
pr = ia->ia6_ndpr;
in6_purgeaddr(&ia->ia_ifa);
if (pr != NULL && pr->ndpr_addrcnt == 0) {
ND6_WLOCK();
nd6_prefix_unlink(pr, NULL);
ND6_WUNLOCK();
nd6_prefix_del(pr);
}
}
static void
in6_unlink_ifa(struct in6_ifaddr *ia, struct ifnet *ifp)
{
char ip6buf[INET6_ADDRSTRLEN];
int remove_lle;
IF_ADDR_WLOCK(ifp);
CK_STAILQ_REMOVE(&ifp->if_addrhead, &ia->ia_ifa, ifaddr, ifa_link);
IF_ADDR_WUNLOCK(ifp);
ifa_free(&ia->ia_ifa); /* if_addrhead */
/*
* Defer the release of what might be the last reference to the
* in6_ifaddr so that it can't be freed before the remainder of the
* cleanup.
*/
IN6_IFADDR_WLOCK();
CK_STAILQ_REMOVE(&V_in6_ifaddrhead, ia, in6_ifaddr, ia_link);
CK_LIST_REMOVE(ia, ia6_hash);
IN6_IFADDR_WUNLOCK();
/*
* Release the reference to the base prefix. There should be a
* positive reference.
*/
remove_lle = 0;
if (ia->ia6_ndpr == NULL) {
nd6log((LOG_NOTICE,
"in6_unlink_ifa: autoconf'ed address "
"%s has no prefix\n", ip6_sprintf(ip6buf, IA6_IN6(ia))));
} else {
ia->ia6_ndpr->ndpr_addrcnt--;
/* Do not delete lles within prefix if refcont != 0 */
if (ia->ia6_ndpr->ndpr_addrcnt == 0)
remove_lle = 1;
ia->ia6_ndpr = NULL;
}
nd6_rem_ifa_lle(ia, remove_lle);
/*
* Also, if the address being removed is autoconf'ed, call
* pfxlist_onlink_check() since the release might affect the status of
* other (detached) addresses.
*/
if ((ia->ia6_flags & IN6_IFF_AUTOCONF)) {
pfxlist_onlink_check();
}
ifa_free(&ia->ia_ifa); /* in6_ifaddrhead */
}
/*
* Notifies other subsystems about address change/arrival:
* 1) Notifies device handler on the first IPv6 address assignment
* 2) Handle routing table changes for P2P links and route
* 3) Handle routing table changes for address host route
*/
static int
in6_notify_ifa(struct ifnet *ifp, struct in6_ifaddr *ia,
struct in6_aliasreq *ifra, int hostIsNew)
{
int error = 0, ifacount = 0;
struct ifaddr *ifa;
struct sockaddr_in6 *pdst;
char ip6buf[INET6_ADDRSTRLEN];
/*
* Give the interface a chance to initialize
* if this is its first address,
*/
if (hostIsNew != 0) {
struct epoch_tracker et;
NET_EPOCH_ENTER(et);
CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
if (ifa->ifa_addr->sa_family != AF_INET6)
continue;
ifacount++;
}
NET_EPOCH_EXIT(et);
}
if (ifacount <= 1 && ifp->if_ioctl) {
error = (*ifp->if_ioctl)(ifp, SIOCSIFADDR, (caddr_t)ia);
if (error)
goto done;
}
/*
* If a new destination address is specified, scrub the old one and
* install the new destination. Note that the interface must be
* p2p or loopback.
*/
pdst = &ifra->ifra_dstaddr;
if (pdst->sin6_family == AF_INET6 &&
!IN6_ARE_ADDR_EQUAL(&pdst->sin6_addr, &ia->ia_dstaddr.sin6_addr)) {
if ((ia->ia_flags & IFA_ROUTE) != 0 &&
(in6_handle_dstaddr_rtrequest(RTM_DELETE, ia) != 0)) {
nd6log((LOG_ERR, "in6_update_ifa_internal: failed to "
"remove a route to the old destination: %s\n",
ip6_sprintf(ip6buf, &ia->ia_addr.sin6_addr)));
/* proceed anyway... */
} else
ia->ia_flags &= ~IFA_ROUTE;
ia->ia_dstaddr = *pdst;
}
/*
* If a new destination address is specified for a point-to-point
* interface, install a route to the destination as an interface
* direct route.
* XXX: the logic below rejects assigning multiple addresses on a p2p
* interface that share the same destination.
*/
if (!(ia->ia_flags & IFA_ROUTE) && ifa_is_p2p(ia)) {
error = in6_handle_dstaddr_rtrequest(RTM_ADD, ia);
if (error)
goto done;
ia->ia_flags |= IFA_ROUTE;
}
/*
* add a loopback route to self if not exists
*/
if (!(ia->ia_flags & IFA_RTSELF) && V_nd6_useloopback) {
error = ifa_add_loopback_route((struct ifaddr *)ia,
(struct sockaddr *)&ia->ia_addr);
if (error == 0)
ia->ia_flags |= IFA_RTSELF;
}
done:
WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL,
"Invoking IPv6 network device address event may sleep");
ifa_ref(&ia->ia_ifa);
EVENTHANDLER_INVOKE(ifaddr_event_ext, ifp, &ia->ia_ifa,
IFADDR_EVENT_ADD);
ifa_free(&ia->ia_ifa);
return (error);
}
/*
* Find an IPv6 interface link-local address specific to an interface.
* ifaddr is returned referenced.
*/
struct in6_ifaddr *
in6ifa_ifpforlinklocal(struct ifnet *ifp, int ignoreflags)
{
struct ifaddr *ifa;
NET_EPOCH_ASSERT();
CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
if (ifa->ifa_addr->sa_family != AF_INET6)
continue;
if (IN6_IS_ADDR_LINKLOCAL(IFA_IN6(ifa))) {
if ((((struct in6_ifaddr *)ifa)->ia6_flags &
ignoreflags) != 0)
continue;
ifa_ref(ifa);
break;
}
}
return ((struct in6_ifaddr *)ifa);
}
/*
* find the interface address corresponding to a given IPv6 address.
* ifaddr is returned referenced if @referenced flag is set.
*/
struct in6_ifaddr *
in6ifa_ifwithaddr(const struct in6_addr *addr, uint32_t zoneid, bool referenced)
{
struct rm_priotracker in6_ifa_tracker;
struct in6_ifaddr *ia;
IN6_IFADDR_RLOCK(&in6_ifa_tracker);
CK_LIST_FOREACH(ia, IN6ADDR_HASH(addr), ia6_hash) {
if (IN6_ARE_ADDR_EQUAL(IA6_IN6(ia), addr)) {
if (zoneid != 0 &&
zoneid != ia->ia_addr.sin6_scope_id)
continue;
if (referenced)
ifa_ref(&ia->ia_ifa);
break;
}
}
IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
return (ia);
}
/*
* find the internet address corresponding to a given interface and address.
* ifaddr is returned referenced.
*/
struct in6_ifaddr *
in6ifa_ifpwithaddr(struct ifnet *ifp, const struct in6_addr *addr)
{
struct epoch_tracker et;
struct ifaddr *ifa;
NET_EPOCH_ENTER(et);
CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
if (ifa->ifa_addr->sa_family != AF_INET6)
continue;
if (IN6_ARE_ADDR_EQUAL(addr, IFA_IN6(ifa))) {
ifa_ref(ifa);
break;
}
}
NET_EPOCH_EXIT(et);
return ((struct in6_ifaddr *)ifa);
}
/*
* Find a link-local scoped address on ifp and return it if any.
*/
struct in6_ifaddr *
in6ifa_llaonifp(struct ifnet *ifp)
{
struct epoch_tracker et;
struct sockaddr_in6 *sin6;
struct ifaddr *ifa;
if (ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED)
return (NULL);
NET_EPOCH_ENTER(et);
CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
if (ifa->ifa_addr->sa_family != AF_INET6)
continue;
sin6 = (struct sockaddr_in6 *)ifa->ifa_addr;
if (IN6_IS_SCOPE_LINKLOCAL(&sin6->sin6_addr) ||
IN6_IS_ADDR_MC_INTFACELOCAL(&sin6->sin6_addr) ||
IN6_IS_ADDR_MC_NODELOCAL(&sin6->sin6_addr))
break;
}
NET_EPOCH_EXIT(et);
return ((struct in6_ifaddr *)ifa);
}
/*
* Convert IP6 address to printable (loggable) representation. Caller
* has to make sure that ip6buf is at least INET6_ADDRSTRLEN long.
*/
static char digits[] = "0123456789abcdef";
char *
ip6_sprintf(char *ip6buf, const struct in6_addr *addr)
{
int i, cnt = 0, maxcnt = 0, idx = 0, index = 0;
char *cp;
const u_int16_t *a = (const u_int16_t *)addr;
const u_int8_t *d;
int dcolon = 0, zero = 0;
cp = ip6buf;
for (i = 0; i < 8; i++) {
if (*(a + i) == 0) {
cnt++;
if (cnt == 1)
idx = i;
}
else if (maxcnt < cnt) {
maxcnt = cnt;
index = idx;
cnt = 0;
}
}
if (maxcnt < cnt) {
maxcnt = cnt;
index = idx;
}
for (i = 0; i < 8; i++) {
if (dcolon == 1) {
if (*a == 0) {
if (i == 7)
*cp++ = ':';
a++;
continue;
} else
dcolon = 2;
}
if (*a == 0) {
if (dcolon == 0 && *(a + 1) == 0 && i == index) {
if (i == 0)
*cp++ = ':';
*cp++ = ':';
dcolon = 1;
} else {
*cp++ = '0';
*cp++ = ':';
}
a++;
continue;
}
d = (const u_char *)a;
/* Try to eliminate leading zeros in printout like in :0001. */
zero = 1;
*cp = digits[*d >> 4];
if (*cp != '0') {
zero = 0;
cp++;
}
*cp = digits[*d++ & 0xf];
if (zero == 0 || (*cp != '0')) {
zero = 0;
cp++;
}
*cp = digits[*d >> 4];
if (zero == 0 || (*cp != '0')) {
zero = 0;
cp++;
}
*cp++ = digits[*d & 0xf];
*cp++ = ':';
a++;
}
*--cp = '\0';
return (ip6buf);
}
int
in6_localaddr(struct in6_addr *in6)
{
struct rm_priotracker in6_ifa_tracker;
struct in6_ifaddr *ia;
if (IN6_IS_ADDR_LOOPBACK(in6) || IN6_IS_ADDR_LINKLOCAL(in6))
return 1;
IN6_IFADDR_RLOCK(&in6_ifa_tracker);
CK_STAILQ_FOREACH(ia, &V_in6_ifaddrhead, ia_link) {
if (IN6_ARE_MASKED_ADDR_EQUAL(in6, &ia->ia_addr.sin6_addr,
&ia->ia_prefixmask.sin6_addr)) {
IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
return 1;
}
}
IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
return (0);
}
/*
* Return 1 if an internet address is for the local host and configured
* on one of its interfaces.
*/
int
in6_localip(struct in6_addr *in6)
{
struct rm_priotracker in6_ifa_tracker;
struct in6_ifaddr *ia;
IN6_IFADDR_RLOCK(&in6_ifa_tracker);
CK_LIST_FOREACH(ia, IN6ADDR_HASH(in6), ia6_hash) {
if (IN6_ARE_ADDR_EQUAL(in6, &ia->ia_addr.sin6_addr)) {
IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
return (1);
}
}
IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
return (0);
}
/*
* Return 1 if an internet address is configured on an interface.
*/
int
in6_ifhasaddr(struct ifnet *ifp, struct in6_addr *addr)
{
struct in6_addr in6;
struct ifaddr *ifa;
struct in6_ifaddr *ia6;
NET_EPOCH_ASSERT();
in6 = *addr;
if (in6_clearscope(&in6))
return (0);
in6_setscope(&in6, ifp, NULL);
CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
if (ifa->ifa_addr->sa_family != AF_INET6)
continue;
ia6 = (struct in6_ifaddr *)ifa;
if (IN6_ARE_ADDR_EQUAL(&ia6->ia_addr.sin6_addr, &in6))
return (1);
}
return (0);
}
int
in6_is_addr_deprecated(struct sockaddr_in6 *sa6)
{
struct rm_priotracker in6_ifa_tracker;
struct in6_ifaddr *ia;
IN6_IFADDR_RLOCK(&in6_ifa_tracker);
CK_LIST_FOREACH(ia, IN6ADDR_HASH(&sa6->sin6_addr), ia6_hash) {
if (IN6_ARE_ADDR_EQUAL(IA6_IN6(ia), &sa6->sin6_addr)) {
if (ia->ia6_flags & IN6_IFF_DEPRECATED) {
IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
return (1); /* true */
}
break;
}
}
IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
return (0); /* false */
}
/*
* return length of part which dst and src are equal
* hard coding...
*/
int
in6_matchlen(struct in6_addr *src, struct in6_addr *dst)
{
int match = 0;
u_char *s = (u_char *)src, *d = (u_char *)dst;
u_char *lim = s + 16, r;
while (s < lim)
if ((r = (*d++ ^ *s++)) != 0) {
while (r < 128) {
match++;
r <<= 1;
}
break;
} else
match += 8;
return match;
}
/* XXX: to be scope conscious */
int
in6_are_prefix_equal(struct in6_addr *p1, struct in6_addr *p2, int len)
{
int bytelen, bitlen;
/* sanity check */
if (0 > len || len > 128) {
log(LOG_ERR, "in6_are_prefix_equal: invalid prefix length(%d)\n",
len);
return (0);
}
bytelen = len / 8;
bitlen = len % 8;
if (bcmp(&p1->s6_addr, &p2->s6_addr, bytelen))
return (0);
if (bitlen != 0 &&
p1->s6_addr[bytelen] >> (8 - bitlen) !=
p2->s6_addr[bytelen] >> (8 - bitlen))
return (0);
return (1);
}
void
in6_prefixlen2mask(struct in6_addr *maskp, int len)
{
u_char maskarray[8] = {0x80, 0xc0, 0xe0, 0xf0, 0xf8, 0xfc, 0xfe, 0xff};
int bytelen, bitlen, i;
/* sanity check */
if (0 > len || len > 128) {
log(LOG_ERR, "in6_prefixlen2mask: invalid prefix length(%d)\n",
len);
return;
}
bzero(maskp, sizeof(*maskp));
bytelen = len / 8;
bitlen = len % 8;
for (i = 0; i < bytelen; i++)
maskp->s6_addr[i] = 0xff;
if (bitlen)
maskp->s6_addr[bytelen] = maskarray[bitlen - 1];
}
/*
* return the best address out of the same scope. if no address was
* found, return the first valid address from designated IF.
*/
struct in6_ifaddr *
in6_ifawithifp(struct ifnet *ifp, struct in6_addr *dst)
{
int dst_scope = in6_addrscope(dst), blen = -1, tlen;
struct ifaddr *ifa;
struct in6_ifaddr *besta = NULL;
struct in6_ifaddr *dep[2]; /* last-resort: deprecated */
NET_EPOCH_ASSERT();
dep[0] = dep[1] = NULL;
/*
* We first look for addresses in the same scope.
* If there is one, return it.
* If two or more, return one which matches the dst longest.
* If none, return one of global addresses assigned other ifs.
*/
CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
if (ifa->ifa_addr->sa_family != AF_INET6)
continue;
if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_ANYCAST)
continue; /* XXX: is there any case to allow anycast? */
if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_NOTREADY)
continue; /* don't use this interface */
if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DETACHED)
continue;
if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DEPRECATED) {
if (V_ip6_use_deprecated)
dep[0] = (struct in6_ifaddr *)ifa;
continue;
}
if (dst_scope == in6_addrscope(IFA_IN6(ifa))) {
/*
* call in6_matchlen() as few as possible
*/
if (besta) {
if (blen == -1)
blen = in6_matchlen(&besta->ia_addr.sin6_addr, dst);
tlen = in6_matchlen(IFA_IN6(ifa), dst);
if (tlen > blen) {
blen = tlen;
besta = (struct in6_ifaddr *)ifa;
}
} else
besta = (struct in6_ifaddr *)ifa;
}
}
if (besta)
return (besta);
CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
if (ifa->ifa_addr->sa_family != AF_INET6)
continue;
if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_ANYCAST)
continue; /* XXX: is there any case to allow anycast? */
if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_NOTREADY)
continue; /* don't use this interface */
if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DETACHED)
continue;
if (((struct in6_ifaddr *)ifa)->ia6_flags & IN6_IFF_DEPRECATED) {
if (V_ip6_use_deprecated)
dep[1] = (struct in6_ifaddr *)ifa;
continue;
}
return (struct in6_ifaddr *)ifa;
}
/* use the last-resort values, that are, deprecated addresses */
if (dep[0])
return dep[0];
if (dep[1])
return dep[1];
return NULL;
}
/*
* perform DAD when interface becomes IFF_UP.
*/
void
in6_if_up(struct ifnet *ifp)
{
struct epoch_tracker et;
struct ifaddr *ifa;
struct in6_ifaddr *ia;
NET_EPOCH_ENTER(et);
CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
if (ifa->ifa_addr->sa_family != AF_INET6)
continue;
ia = (struct in6_ifaddr *)ifa;
if (ia->ia6_flags & IN6_IFF_TENTATIVE) {
/*
* The TENTATIVE flag was likely set by hand
* beforehand, implicitly indicating the need for DAD.
* We may be able to skip the random delay in this
* case, but we impose delays just in case.
*/
nd6_dad_start(ifa,
arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz));
}
}
NET_EPOCH_EXIT(et);
/*
* special cases, like 6to4, are handled in in6_ifattach
*/
in6_ifattach(ifp, NULL);
}
int
in6if_do_dad(struct ifnet *ifp)
{
if ((ifp->if_flags & IFF_LOOPBACK) != 0)
return (0);
if ((ifp->if_flags & IFF_MULTICAST) == 0)
return (0);
if ((ND_IFINFO(ifp)->flags &
(ND6_IFF_IFDISABLED | ND6_IFF_NO_DAD)) != 0)
return (0);
return (1);
}
/*
* Calculate max IPv6 MTU through all the interfaces and store it
* to in6_maxmtu.
*/
void
in6_setmaxmtu(void)
{
struct epoch_tracker et;
unsigned long maxmtu = 0;
struct ifnet *ifp;
NET_EPOCH_ENTER(et);
CK_STAILQ_FOREACH(ifp, &V_ifnet, if_link) {
/* this function can be called during ifnet initialization */
if (!ifp->if_afdata[AF_INET6])
continue;
if ((ifp->if_flags & IFF_LOOPBACK) == 0 &&
IN6_LINKMTU(ifp) > maxmtu)
maxmtu = IN6_LINKMTU(ifp);
}
NET_EPOCH_EXIT(et);
if (maxmtu) /* update only when maxmtu is positive */
V_in6_maxmtu = maxmtu;
}
/*
* Provide the length of interface identifiers to be used for the link attached
* to the given interface. The length should be defined in "IPv6 over
* xxx-link" document. Note that address architecture might also define
* the length for a particular set of address prefixes, regardless of the
* link type. As clarified in rfc2462bis, those two definitions should be
* consistent, and those really are as of August 2004.
*/
int
in6_if2idlen(struct ifnet *ifp)
{
switch (ifp->if_type) {
case IFT_ETHER: /* RFC2464 */
case IFT_PROPVIRTUAL: /* XXX: no RFC. treat it as ether */
case IFT_L2VLAN: /* ditto */
case IFT_BRIDGE: /* bridge(4) only does Ethernet-like links */
case IFT_INFINIBAND:
return (64);
case IFT_PPP: /* RFC2472 */
return (64);
case IFT_FRELAY: /* RFC2590 */
return (64);
case IFT_IEEE1394: /* RFC3146 */
return (64);
case IFT_GIF:
return (64); /* draft-ietf-v6ops-mech-v2-07 */
case IFT_LOOP:
return (64); /* XXX: is this really correct? */
default:
/*
* Unknown link type:
* It might be controversial to use the today's common constant
* of 64 for these cases unconditionally. For full compliance,
* we should return an error in this case. On the other hand,
* if we simply miss the standard for the link type or a new
* standard is defined for a new link type, the IFID length
* is very likely to be the common constant. As a compromise,
* we always use the constant, but make an explicit notice
* indicating the "unknown" case.
*/
printf("in6_if2idlen: unknown link type (%d)\n", ifp->if_type);
return (64);
}
}
struct in6_llentry {
struct llentry base;
};
#define IN6_LLTBL_DEFAULT_HSIZE 32
#define IN6_LLTBL_HASH(k, h) \
(((((((k >> 8) ^ k) >> 8) ^ k) >> 8) ^ k) & ((h) - 1))
/*
* Do actual deallocation of @lle.
*/
static void
in6_lltable_destroy_lle_unlocked(epoch_context_t ctx)
{
struct llentry *lle;
lle = __containerof(ctx, struct llentry, lle_epoch_ctx);
LLE_LOCK_DESTROY(lle);
LLE_REQ_DESTROY(lle);
free(lle, M_LLTABLE);
}
/*
* Called by LLE_FREE_LOCKED when number of references
* drops to zero.
*/
static void
in6_lltable_destroy_lle(struct llentry *lle)
{
LLE_WUNLOCK(lle);
NET_EPOCH_CALL(in6_lltable_destroy_lle_unlocked, &lle->lle_epoch_ctx);
}
static struct llentry *
in6_lltable_new(const struct in6_addr *addr6, u_int flags)
{
struct in6_llentry *lle;
lle = malloc(sizeof(struct in6_llentry), M_LLTABLE, M_NOWAIT | M_ZERO);
if (lle == NULL) /* NB: caller generates msg */
return NULL;
lle->base.r_l3addr.addr6 = *addr6;
lle->base.lle_refcnt = 1;
lle->base.lle_free = in6_lltable_destroy_lle;
LLE_LOCK_INIT(&lle->base);
LLE_REQ_INIT(&lle->base);
callout_init(&lle->base.lle_timer, 1);
return (&lle->base);
}
static int
in6_lltable_match_prefix(const struct sockaddr *saddr,
const struct sockaddr *smask, u_int flags, struct llentry *lle)
{
const struct in6_addr *addr, *mask, *lle_addr;
addr = &((const struct sockaddr_in6 *)saddr)->sin6_addr;
mask = &((const struct sockaddr_in6 *)smask)->sin6_addr;
lle_addr = &lle->r_l3addr.addr6;
if (IN6_ARE_MASKED_ADDR_EQUAL(lle_addr, addr, mask) == 0)
return (0);
if (lle->la_flags & LLE_IFADDR) {
/*
* Delete LLE_IFADDR records IFF address & flag matches.
* Note that addr is the interface address within prefix
* being matched.
*/
if (IN6_ARE_ADDR_EQUAL(addr, lle_addr) &&
(flags & LLE_STATIC) != 0)
return (1);
return (0);
}
/* flags & LLE_STATIC means deleting both dynamic and static entries */
if ((flags & LLE_STATIC) || !(lle->la_flags & LLE_STATIC))
return (1);
return (0);
}
static void
in6_lltable_free_entry(struct lltable *llt, struct llentry *lle)
{
struct ifnet *ifp;
LLE_WLOCK_ASSERT(lle);
KASSERT(llt != NULL, ("lltable is NULL"));
/* Unlink entry from table */
if ((lle->la_flags & LLE_LINKED) != 0) {
ifp = llt->llt_ifp;
IF_AFDATA_WLOCK_ASSERT(ifp);
lltable_unlink_entry(llt, lle);
}
llentry_free(lle);
}
static int
in6_lltable_rtcheck(struct ifnet *ifp,
u_int flags,
const struct sockaddr *l3addr)
{
const struct sockaddr_in6 *sin6;
struct nhop_object *nh;
struct in6_addr dst;
uint32_t scopeid;
char ip6buf[INET6_ADDRSTRLEN];
int fibnum;
NET_EPOCH_ASSERT();
KASSERT(l3addr->sa_family == AF_INET6,
("sin_family %d", l3addr->sa_family));
sin6 = (const struct sockaddr_in6 *)l3addr;
in6_splitscope(&sin6->sin6_addr, &dst, &scopeid);
fibnum = V_rt_add_addr_allfibs ? RT_DEFAULT_FIB : ifp->if_fib;
nh = fib6_lookup(fibnum, &dst, scopeid, NHR_NONE, 0);
if (nh && ((nh->nh_flags & NHF_GATEWAY) || nh->nh_ifp != ifp)) {
struct ifaddr *ifa;
/*
* Create an ND6 cache for an IPv6 neighbor
* that is not covered by our own prefix.
*/
ifa = ifaof_ifpforaddr(l3addr, ifp);
if (ifa != NULL) {
return 0;
}
log(LOG_INFO, "IPv6 address: \"%s\" is not on the network\n",
ip6_sprintf(ip6buf, &sin6->sin6_addr));
return EINVAL;
}
return 0;
}
static inline uint32_t
in6_lltable_hash_dst(const struct in6_addr *dst, uint32_t hsize)
{
return (IN6_LLTBL_HASH(dst->s6_addr32[3], hsize));
}
static uint32_t
in6_lltable_hash(const struct llentry *lle, uint32_t hsize)
{
return (in6_lltable_hash_dst(&lle->r_l3addr.addr6, hsize));
}
static void
in6_lltable_fill_sa_entry(const struct llentry *lle, struct sockaddr *sa)
{
struct sockaddr_in6 *sin6;
sin6 = (struct sockaddr_in6 *)sa;
bzero(sin6, sizeof(*sin6));
sin6->sin6_family = AF_INET6;
sin6->sin6_len = sizeof(*sin6);
sin6->sin6_addr = lle->r_l3addr.addr6;
}
static inline struct llentry *
in6_lltable_find_dst(struct lltable *llt, const struct in6_addr *dst)
{
struct llentry *lle;
struct llentries *lleh;
u_int hashidx;
hashidx = in6_lltable_hash_dst(dst, llt->llt_hsize);
lleh = &llt->lle_head[hashidx];
CK_LIST_FOREACH(lle, lleh, lle_next) {
if (lle->la_flags & LLE_DELETED)
continue;
if (IN6_ARE_ADDR_EQUAL(&lle->r_l3addr.addr6, dst))
break;
}
return (lle);
}
static void
in6_lltable_delete_entry(struct lltable *llt, struct llentry *lle)
{
lle->la_flags |= LLE_DELETED;
EVENTHANDLER_INVOKE(lle_event, lle, LLENTRY_DELETED);
#ifdef DIAGNOSTIC
log(LOG_INFO, "ifaddr cache = %p is deleted\n", lle);
#endif
llentry_free(lle);
}
static struct llentry *
in6_lltable_alloc(struct lltable *llt, u_int flags,
const struct sockaddr *l3addr)
{
const struct sockaddr_in6 *sin6 = (const struct sockaddr_in6 *)l3addr;
struct ifnet *ifp = llt->llt_ifp;
struct llentry *lle;
char linkhdr[LLE_MAX_LINKHDR];
size_t linkhdrsize;
int lladdr_off;
KASSERT(l3addr->sa_family == AF_INET6,
("sin_family %d", l3addr->sa_family));
/*
* A route that covers the given address must have
* been installed 1st because we are doing a resolution,
* verify this.
*/
if (!(flags & LLE_IFADDR) &&
in6_lltable_rtcheck(ifp, flags, l3addr) != 0)
return (NULL);
lle = in6_lltable_new(&sin6->sin6_addr, flags);
if (lle == NULL) {
log(LOG_INFO, "lla_lookup: new lle malloc failed\n");
return (NULL);
}
lle->la_flags = flags;
if ((flags & LLE_IFADDR) == LLE_IFADDR) {
linkhdrsize = LLE_MAX_LINKHDR;
if (lltable_calc_llheader(ifp, AF_INET6, IF_LLADDR(ifp),
linkhdr, &linkhdrsize, &lladdr_off) != 0) {
- NET_EPOCH_CALL(in6_lltable_destroy_lle_unlocked, &lle->lle_epoch_ctx);
+ in6_lltable_free_entry(llt, lle);
return (NULL);
}
lltable_set_entry_addr(ifp, lle, linkhdr, linkhdrsize,
lladdr_off);
lle->la_flags |= LLE_STATIC;
}
if ((lle->la_flags & LLE_STATIC) != 0)
lle->ln_state = ND6_LLINFO_REACHABLE;
return (lle);
}
static struct llentry *
in6_lltable_lookup(struct lltable *llt, u_int flags,
const struct sockaddr *l3addr)
{
const struct sockaddr_in6 *sin6 = (const struct sockaddr_in6 *)l3addr;
int family = flags >> 16;
struct llentry *lle;
IF_AFDATA_LOCK_ASSERT(llt->llt_ifp);
KASSERT(l3addr->sa_family == AF_INET6,
("sin_family %d", l3addr->sa_family));
KASSERT((flags & (LLE_UNLOCKED | LLE_EXCLUSIVE)) !=
(LLE_UNLOCKED | LLE_EXCLUSIVE),
("wrong lle request flags: %#x", flags));
lle = in6_lltable_find_dst(llt, &sin6->sin6_addr);
if (__predict_false(family != AF_INET6))
lle = llentry_lookup_family(lle, family);
if (lle == NULL)
return (NULL);
if (flags & LLE_UNLOCKED)
return (lle);
if (flags & LLE_EXCLUSIVE)
LLE_WLOCK(lle);
else
LLE_RLOCK(lle);
/*
* If the afdata lock is not held, the LLE may have been unlinked while
* we were blocked on the LLE lock. Check for this case.
*/
if (__predict_false((lle->la_flags & LLE_LINKED) == 0)) {
if (flags & LLE_EXCLUSIVE)
LLE_WUNLOCK(lle);
else
LLE_RUNLOCK(lle);
return (NULL);
}
return (lle);
}
static int
in6_lltable_dump_entry(struct lltable *llt, struct llentry *lle,
struct sysctl_req *wr)
{
struct ifnet *ifp = llt->llt_ifp;
/* XXX stack use */
struct {
struct rt_msghdr rtm;
struct sockaddr_in6 sin6;
/*
* ndp.c assumes that sdl is word aligned
*/
#ifdef __LP64__
uint32_t pad;
#endif
struct sockaddr_dl sdl;
} ndpc;
struct sockaddr_dl *sdl;
int error;
bzero(&ndpc, sizeof(ndpc));
/* skip deleted entries */
if ((lle->la_flags & LLE_DELETED) == LLE_DELETED)
return (0);
/* Skip if jailed and not a valid IP of the prison. */
lltable_fill_sa_entry(lle, (struct sockaddr *)&ndpc.sin6);
if (prison_if(wr->td->td_ucred, (struct sockaddr *)&ndpc.sin6) != 0)
return (0);
/*
* produce a msg made of:
* struct rt_msghdr;
* struct sockaddr_in6 (IPv6)
* struct sockaddr_dl;
*/
ndpc.rtm.rtm_msglen = sizeof(ndpc);
ndpc.rtm.rtm_version = RTM_VERSION;
ndpc.rtm.rtm_type = RTM_GET;
ndpc.rtm.rtm_flags = RTF_UP;
ndpc.rtm.rtm_addrs = RTA_DST | RTA_GATEWAY;
sa6_recoverscope(&ndpc.sin6);
/* publish */
if (lle->la_flags & LLE_PUB)
ndpc.rtm.rtm_flags |= RTF_ANNOUNCE;
sdl = &ndpc.sdl;
sdl->sdl_family = AF_LINK;
sdl->sdl_len = sizeof(*sdl);
sdl->sdl_index = ifp->if_index;
sdl->sdl_type = ifp->if_type;
if ((lle->la_flags & LLE_VALID) == LLE_VALID) {
sdl->sdl_alen = ifp->if_addrlen;
bcopy(lle->ll_addr, LLADDR(sdl), ifp->if_addrlen);
} else {
sdl->sdl_alen = 0;
bzero(LLADDR(sdl), ifp->if_addrlen);
}
if (lle->la_expire != 0)
ndpc.rtm.rtm_rmx.rmx_expire = lle->la_expire +
lle->lle_remtime / hz + time_second - time_uptime;
ndpc.rtm.rtm_flags |= (RTF_HOST | RTF_LLDATA);
if (lle->la_flags & LLE_STATIC)
ndpc.rtm.rtm_flags |= RTF_STATIC;
if (lle->la_flags & LLE_IFADDR)
ndpc.rtm.rtm_flags |= RTF_PINNED;
if (lle->ln_router != 0)
ndpc.rtm.rtm_flags |= RTF_GATEWAY;
ndpc.rtm.rtm_rmx.rmx_pksent = lle->la_asked;
/* Store state in rmx_weight value */
ndpc.rtm.rtm_rmx.rmx_state = lle->ln_state;
ndpc.rtm.rtm_index = ifp->if_index;
error = SYSCTL_OUT(wr, &ndpc, sizeof(ndpc));
return (error);
}
static struct lltable *
in6_lltattach(struct ifnet *ifp)
{
struct lltable *llt;
llt = lltable_allocate_htbl(IN6_LLTBL_DEFAULT_HSIZE);
llt->llt_af = AF_INET6;
llt->llt_ifp = ifp;
llt->llt_lookup = in6_lltable_lookup;
llt->llt_alloc_entry = in6_lltable_alloc;
llt->llt_delete_entry = in6_lltable_delete_entry;
llt->llt_dump_entry = in6_lltable_dump_entry;
llt->llt_hash = in6_lltable_hash;
llt->llt_fill_sa_entry = in6_lltable_fill_sa_entry;
llt->llt_free_entry = in6_lltable_free_entry;
llt->llt_match_prefix = in6_lltable_match_prefix;
llt->llt_mark_used = llentry_mark_used;
lltable_link(llt);
return (llt);
}
struct lltable *
in6_lltable_get(struct ifnet *ifp)
{
struct lltable *llt = NULL;
void *afdata_ptr = ifp->if_afdata[AF_INET6];
if (afdata_ptr != NULL)
llt = ((struct in6_ifextra *)afdata_ptr)->lltable;
return (llt);
}
void *
in6_domifattach(struct ifnet *ifp)
{
struct in6_ifextra *ext;
/* There are not IPv6-capable interfaces. */
switch (ifp->if_type) {
case IFT_PFLOG:
case IFT_PFSYNC:
case IFT_USB:
return (NULL);
}
ext = (struct in6_ifextra *)malloc(sizeof(*ext), M_IFADDR, M_WAITOK);
bzero(ext, sizeof(*ext));
ext->in6_ifstat = malloc(sizeof(counter_u64_t) *
sizeof(struct in6_ifstat) / sizeof(uint64_t), M_IFADDR, M_WAITOK);
COUNTER_ARRAY_ALLOC(ext->in6_ifstat,
sizeof(struct in6_ifstat) / sizeof(uint64_t), M_WAITOK);
ext->icmp6_ifstat = malloc(sizeof(counter_u64_t) *
sizeof(struct icmp6_ifstat) / sizeof(uint64_t), M_IFADDR,
M_WAITOK);
COUNTER_ARRAY_ALLOC(ext->icmp6_ifstat,
sizeof(struct icmp6_ifstat) / sizeof(uint64_t), M_WAITOK);
ext->nd_ifinfo = nd6_ifattach(ifp);
ext->scope6_id = scope6_ifattach(ifp);
ext->lltable = in6_lltattach(ifp);
ext->mld_ifinfo = mld_domifattach(ifp);
return ext;
}
int
in6_domifmtu(struct ifnet *ifp)
{
if (ifp->if_afdata[AF_INET6] == NULL)
return ifp->if_mtu;
return (IN6_LINKMTU(ifp));
}
void
in6_domifdetach(struct ifnet *ifp, void *aux)
{
struct in6_ifextra *ext = (struct in6_ifextra *)aux;
mld_domifdetach(ifp);
scope6_ifdetach(ext->scope6_id);
nd6_ifdetach(ifp, ext->nd_ifinfo);
lltable_free(ext->lltable);
COUNTER_ARRAY_FREE(ext->in6_ifstat,
sizeof(struct in6_ifstat) / sizeof(uint64_t));
free(ext->in6_ifstat, M_IFADDR);
COUNTER_ARRAY_FREE(ext->icmp6_ifstat,
sizeof(struct icmp6_ifstat) / sizeof(uint64_t));
free(ext->icmp6_ifstat, M_IFADDR);
free(ext, M_IFADDR);
}
/*
* Convert sockaddr_in6 to sockaddr_in. Original sockaddr_in6 must be
* v4 mapped addr or v4 compat addr
*/
void
in6_sin6_2_sin(struct sockaddr_in *sin, struct sockaddr_in6 *sin6)
{
bzero(sin, sizeof(*sin));
sin->sin_len = sizeof(struct sockaddr_in);
sin->sin_family = AF_INET;
sin->sin_port = sin6->sin6_port;
sin->sin_addr.s_addr = sin6->sin6_addr.s6_addr32[3];
}
/* Convert sockaddr_in to sockaddr_in6 in v4 mapped addr format. */
void
in6_sin_2_v4mapsin6(struct sockaddr_in *sin, struct sockaddr_in6 *sin6)
{
bzero(sin6, sizeof(*sin6));
sin6->sin6_len = sizeof(struct sockaddr_in6);
sin6->sin6_family = AF_INET6;
sin6->sin6_port = sin->sin_port;
sin6->sin6_addr.s6_addr32[0] = 0;
sin6->sin6_addr.s6_addr32[1] = 0;
sin6->sin6_addr.s6_addr32[2] = IPV6_ADDR_INT32_SMP;
sin6->sin6_addr.s6_addr32[3] = sin->sin_addr.s_addr;
}
/* Convert sockaddr_in6 into sockaddr_in. */
void
in6_sin6_2_sin_in_sock(struct sockaddr *nam)
{
struct sockaddr_in *sin_p;
struct sockaddr_in6 sin6;
/*
* Save original sockaddr_in6 addr and convert it
* to sockaddr_in.
*/
sin6 = *(struct sockaddr_in6 *)nam;
sin_p = (struct sockaddr_in *)nam;
in6_sin6_2_sin(sin_p, &sin6);
}
/* Convert sockaddr_in into sockaddr_in6 in v4 mapped addr format. */
void
in6_sin_2_v4mapsin6_in_sock(struct sockaddr **nam)
{
struct sockaddr_in *sin_p;
struct sockaddr_in6 *sin6_p;
sin6_p = malloc(sizeof *sin6_p, M_SONAME, M_WAITOK);
sin_p = (struct sockaddr_in *)*nam;
in6_sin_2_v4mapsin6(sin_p, sin6_p);
free(*nam, M_SONAME);
*nam = (struct sockaddr *)sin6_p;
}
diff --git a/sys/netinet6/nd6.c b/sys/netinet6/nd6.c
index 6a9e2a4fdd7c..69da61aa9c5c 100644
--- a/sys/netinet6/nd6.c
+++ b/sys/netinet6/nd6.c
@@ -1,2794 +1,2796 @@
/*-
* SPDX-License-Identifier: BSD-3-Clause
*
* Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the project nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* $KAME: nd6.c,v 1.144 2001/05/24 07:44:00 itojun Exp $
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include "opt_inet.h"
#include "opt_inet6.h"
#include "opt_route.h"
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/eventhandler.h>
#include <sys/callout.h>
#include <sys/lock.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/mutex.h>
#include <sys/socket.h>
#include <sys/sockio.h>
#include <sys/time.h>
#include <sys/kernel.h>
#include <sys/protosw.h>
#include <sys/errno.h>
#include <sys/syslog.h>
#include <sys/rwlock.h>
#include <sys/queue.h>
#include <sys/sdt.h>
#include <sys/sysctl.h>
#include <net/if.h>
#include <net/if_var.h>
#include <net/if_dl.h>
#include <net/if_types.h>
#include <net/route.h>
#include <net/route/route_ctl.h>
#include <net/route/nhop.h>
#include <net/vnet.h>
#include <netinet/in.h>
#include <netinet/in_kdtrace.h>
#include <net/if_llatbl.h>
#include <netinet/if_ether.h>
#include <netinet6/in6_var.h>
#include <netinet/ip6.h>
#include <netinet6/ip6_var.h>
#include <netinet6/scope6_var.h>
#include <netinet6/nd6.h>
#include <netinet6/in6_ifattach.h>
#include <netinet/icmp6.h>
#include <netinet6/send.h>
#include <sys/limits.h>
#include <security/mac/mac_framework.h>
#define ND6_SLOWTIMER_INTERVAL (60 * 60) /* 1 hour */
#define ND6_RECALC_REACHTM_INTERVAL (60 * 120) /* 2 hours */
#define SIN6(s) ((const struct sockaddr_in6 *)(s))
MALLOC_DEFINE(M_IP6NDP, "ip6ndp", "IPv6 Neighbor Discovery");
/* timer values */
VNET_DEFINE(int, nd6_prune) = 1; /* walk list every 1 seconds */
VNET_DEFINE(int, nd6_delay) = 5; /* delay first probe time 5 second */
VNET_DEFINE(int, nd6_umaxtries) = 3; /* maximum unicast query */
VNET_DEFINE(int, nd6_mmaxtries) = 3; /* maximum multicast query */
VNET_DEFINE(int, nd6_useloopback) = 1; /* use loopback interface for
* local traffic */
VNET_DEFINE(int, nd6_gctimer) = (60 * 60 * 24); /* 1 day: garbage
* collection timer */
/* preventing too many loops in ND option parsing */
VNET_DEFINE_STATIC(int, nd6_maxndopt) = 10; /* max # of ND options allowed */
VNET_DEFINE(int, nd6_maxnudhint) = 0; /* max # of subsequent upper
* layer hints */
VNET_DEFINE_STATIC(int, nd6_maxqueuelen) = 16; /* max pkts cached in unresolved
* ND entries */
#define V_nd6_maxndopt VNET(nd6_maxndopt)
#define V_nd6_maxqueuelen VNET(nd6_maxqueuelen)
#ifdef ND6_DEBUG
VNET_DEFINE(int, nd6_debug) = 1;
#else
VNET_DEFINE(int, nd6_debug) = 0;
#endif
static eventhandler_tag lle_event_eh, iflladdr_event_eh, ifnet_link_event_eh;
VNET_DEFINE(struct nd_prhead, nd_prefix);
VNET_DEFINE(struct rwlock, nd6_lock);
VNET_DEFINE(uint64_t, nd6_list_genid);
VNET_DEFINE(struct mtx, nd6_onlink_mtx);
VNET_DEFINE(int, nd6_recalc_reachtm_interval) = ND6_RECALC_REACHTM_INTERVAL;
#define V_nd6_recalc_reachtm_interval VNET(nd6_recalc_reachtm_interval)
int (*send_sendso_input_hook)(struct mbuf *, struct ifnet *, int, int);
static int nd6_is_new_addr_neighbor(const struct sockaddr_in6 *,
struct ifnet *);
static void nd6_setmtu0(struct ifnet *, struct nd_ifinfo *);
static void nd6_slowtimo(void *);
static int regen_tmpaddr(struct in6_ifaddr *);
static void nd6_free(struct llentry **, int);
static void nd6_free_redirect(const struct llentry *);
static void nd6_llinfo_timer(void *);
static void nd6_llinfo_settimer_locked(struct llentry *, long);
static void clear_llinfo_pqueue(struct llentry *);
static int nd6_resolve_slow(struct ifnet *, int, int, struct mbuf *,
const struct sockaddr_in6 *, u_char *, uint32_t *, struct llentry **);
static int nd6_need_cache(struct ifnet *);
VNET_DEFINE_STATIC(struct callout, nd6_slowtimo_ch);
#define V_nd6_slowtimo_ch VNET(nd6_slowtimo_ch)
VNET_DEFINE_STATIC(struct callout, nd6_timer_ch);
#define V_nd6_timer_ch VNET(nd6_timer_ch)
SYSCTL_DECL(_net_inet6_icmp6);
static void
nd6_lle_event(void *arg __unused, struct llentry *lle, int evt)
{
struct rt_addrinfo rtinfo;
struct sockaddr_in6 dst;
struct sockaddr_dl gw;
struct ifnet *ifp;
int type;
int fibnum;
LLE_WLOCK_ASSERT(lle);
if (lltable_get_af(lle->lle_tbl) != AF_INET6)
return;
switch (evt) {
case LLENTRY_RESOLVED:
type = RTM_ADD;
KASSERT(lle->la_flags & LLE_VALID,
("%s: %p resolved but not valid?", __func__, lle));
break;
case LLENTRY_EXPIRED:
type = RTM_DELETE;
break;
default:
return;
}
ifp = lltable_get_ifp(lle->lle_tbl);
bzero(&dst, sizeof(dst));
bzero(&gw, sizeof(gw));
bzero(&rtinfo, sizeof(rtinfo));
lltable_fill_sa_entry(lle, (struct sockaddr *)&dst);
dst.sin6_scope_id = in6_getscopezone(ifp,
in6_addrscope(&dst.sin6_addr));
gw.sdl_len = sizeof(struct sockaddr_dl);
gw.sdl_family = AF_LINK;
gw.sdl_alen = ifp->if_addrlen;
gw.sdl_index = ifp->if_index;
gw.sdl_type = ifp->if_type;
if (evt == LLENTRY_RESOLVED)
bcopy(lle->ll_addr, gw.sdl_data, ifp->if_addrlen);
rtinfo.rti_info[RTAX_DST] = (struct sockaddr *)&dst;
rtinfo.rti_info[RTAX_GATEWAY] = (struct sockaddr *)&gw;
rtinfo.rti_addrs = RTA_DST | RTA_GATEWAY;
fibnum = V_rt_add_addr_allfibs ? RT_ALL_FIBS : ifp->if_fib;
rt_missmsg_fib(type, &rtinfo, RTF_HOST | RTF_LLDATA | (
type == RTM_ADD ? RTF_UP: 0), 0, fibnum);
}
/*
* A handler for interface link layer address change event.
*/
static void
nd6_iflladdr(void *arg __unused, struct ifnet *ifp)
{
if (ifp->if_afdata[AF_INET6] == NULL)
return;
lltable_update_ifaddr(LLTABLE6(ifp));
}
void
nd6_init(void)
{
mtx_init(&V_nd6_onlink_mtx, "nd6 onlink", NULL, MTX_DEF);
rw_init(&V_nd6_lock, "nd6 list");
LIST_INIT(&V_nd_prefix);
nd6_defrouter_init();
/* Start timers. */
callout_init(&V_nd6_slowtimo_ch, 1);
callout_reset(&V_nd6_slowtimo_ch, ND6_SLOWTIMER_INTERVAL * hz,
nd6_slowtimo, curvnet);
callout_init(&V_nd6_timer_ch, 1);
callout_reset(&V_nd6_timer_ch, hz, nd6_timer, curvnet);
nd6_dad_init();
if (IS_DEFAULT_VNET(curvnet)) {
lle_event_eh = EVENTHANDLER_REGISTER(lle_event, nd6_lle_event,
NULL, EVENTHANDLER_PRI_ANY);
iflladdr_event_eh = EVENTHANDLER_REGISTER(iflladdr_event,
nd6_iflladdr, NULL, EVENTHANDLER_PRI_ANY);
ifnet_link_event_eh = EVENTHANDLER_REGISTER(ifnet_link_event,
nd6_ifnet_link_event, NULL, EVENTHANDLER_PRI_ANY);
}
}
#ifdef VIMAGE
void
nd6_destroy()
{
callout_drain(&V_nd6_slowtimo_ch);
callout_drain(&V_nd6_timer_ch);
if (IS_DEFAULT_VNET(curvnet)) {
EVENTHANDLER_DEREGISTER(ifnet_link_event, ifnet_link_event_eh);
EVENTHANDLER_DEREGISTER(lle_event, lle_event_eh);
EVENTHANDLER_DEREGISTER(iflladdr_event, iflladdr_event_eh);
}
rw_destroy(&V_nd6_lock);
mtx_destroy(&V_nd6_onlink_mtx);
}
#endif
struct nd_ifinfo *
nd6_ifattach(struct ifnet *ifp)
{
struct nd_ifinfo *nd;
nd = malloc(sizeof(*nd), M_IP6NDP, M_WAITOK | M_ZERO);
nd->initialized = 1;
nd->chlim = IPV6_DEFHLIM;
nd->basereachable = REACHABLE_TIME;
nd->reachable = ND_COMPUTE_RTIME(nd->basereachable);
nd->retrans = RETRANS_TIMER;
nd->flags = ND6_IFF_PERFORMNUD;
/* Set IPv6 disabled on all interfaces but loopback by default. */
if ((ifp->if_flags & IFF_LOOPBACK) == 0)
nd->flags |= ND6_IFF_IFDISABLED;
/* A loopback interface always has ND6_IFF_AUTO_LINKLOCAL.
* XXXHRS: Clear ND6_IFF_AUTO_LINKLOCAL on an IFT_BRIDGE interface by
* default regardless of the V_ip6_auto_linklocal configuration to
* give a reasonable default behavior.
*/
if ((V_ip6_auto_linklocal && ifp->if_type != IFT_BRIDGE) ||
(ifp->if_flags & IFF_LOOPBACK))
nd->flags |= ND6_IFF_AUTO_LINKLOCAL;
/*
* A loopback interface does not need to accept RTADV.
* XXXHRS: Clear ND6_IFF_ACCEPT_RTADV on an IFT_BRIDGE interface by
* default regardless of the V_ip6_accept_rtadv configuration to
* prevent the interface from accepting RA messages arrived
* on one of the member interfaces with ND6_IFF_ACCEPT_RTADV.
*/
if (V_ip6_accept_rtadv &&
!(ifp->if_flags & IFF_LOOPBACK) &&
(ifp->if_type != IFT_BRIDGE)) {
nd->flags |= ND6_IFF_ACCEPT_RTADV;
/* If we globally accept rtadv, assume IPv6 on. */
nd->flags &= ~ND6_IFF_IFDISABLED;
}
if (V_ip6_no_radr && !(ifp->if_flags & IFF_LOOPBACK))
nd->flags |= ND6_IFF_NO_RADR;
/* XXX: we cannot call nd6_setmtu since ifp is not fully initialized */
nd6_setmtu0(ifp, nd);
return nd;
}
void
nd6_ifdetach(struct ifnet *ifp, struct nd_ifinfo *nd)
{
struct epoch_tracker et;
struct ifaddr *ifa, *next;
NET_EPOCH_ENTER(et);
CK_STAILQ_FOREACH_SAFE(ifa, &ifp->if_addrhead, ifa_link, next) {
if (ifa->ifa_addr->sa_family != AF_INET6)
continue;
/* stop DAD processing */
nd6_dad_stop(ifa);
}
NET_EPOCH_EXIT(et);
free(nd, M_IP6NDP);
}
/*
* Reset ND level link MTU. This function is called when the physical MTU
* changes, which means we might have to adjust the ND level MTU.
*/
void
nd6_setmtu(struct ifnet *ifp)
{
if (ifp->if_afdata[AF_INET6] == NULL)
return;
nd6_setmtu0(ifp, ND_IFINFO(ifp));
}
/* XXX todo: do not maintain copy of ifp->if_mtu in ndi->maxmtu */
void
nd6_setmtu0(struct ifnet *ifp, struct nd_ifinfo *ndi)
{
u_int32_t omaxmtu;
omaxmtu = ndi->maxmtu;
ndi->maxmtu = ifp->if_mtu;
/*
* Decreasing the interface MTU under IPV6 minimum MTU may cause
* undesirable situation. We thus notify the operator of the change
* explicitly. The check for omaxmtu is necessary to restrict the
* log to the case of changing the MTU, not initializing it.
*/
if (omaxmtu >= IPV6_MMTU && ndi->maxmtu < IPV6_MMTU) {
log(LOG_NOTICE, "nd6_setmtu0: "
"new link MTU on %s (%lu) is too small for IPv6\n",
if_name(ifp), (unsigned long)ndi->maxmtu);
}
if (ndi->maxmtu > V_in6_maxmtu)
in6_setmaxmtu(); /* check all interfaces just in case */
}
void
nd6_option_init(void *opt, int icmp6len, union nd_opts *ndopts)
{
bzero(ndopts, sizeof(*ndopts));
ndopts->nd_opts_search = (struct nd_opt_hdr *)opt;
ndopts->nd_opts_last
= (struct nd_opt_hdr *)(((u_char *)opt) + icmp6len);
if (icmp6len == 0) {
ndopts->nd_opts_done = 1;
ndopts->nd_opts_search = NULL;
}
}
/*
* Take one ND option.
*/
struct nd_opt_hdr *
nd6_option(union nd_opts *ndopts)
{
struct nd_opt_hdr *nd_opt;
int olen;
KASSERT(ndopts != NULL, ("%s: ndopts == NULL", __func__));
KASSERT(ndopts->nd_opts_last != NULL, ("%s: uninitialized ndopts",
__func__));
if (ndopts->nd_opts_search == NULL)
return NULL;
if (ndopts->nd_opts_done)
return NULL;
nd_opt = ndopts->nd_opts_search;
/* make sure nd_opt_len is inside the buffer */
if ((caddr_t)&nd_opt->nd_opt_len >= (caddr_t)ndopts->nd_opts_last) {
bzero(ndopts, sizeof(*ndopts));
return NULL;
}
olen = nd_opt->nd_opt_len << 3;
if (olen == 0) {
/*
* Message validation requires that all included
* options have a length that is greater than zero.
*/
bzero(ndopts, sizeof(*ndopts));
return NULL;
}
ndopts->nd_opts_search = (struct nd_opt_hdr *)((caddr_t)nd_opt + olen);
if (ndopts->nd_opts_search > ndopts->nd_opts_last) {
/* option overruns the end of buffer, invalid */
bzero(ndopts, sizeof(*ndopts));
return NULL;
} else if (ndopts->nd_opts_search == ndopts->nd_opts_last) {
/* reached the end of options chain */
ndopts->nd_opts_done = 1;
ndopts->nd_opts_search = NULL;
}
return nd_opt;
}
/*
* Parse multiple ND options.
* This function is much easier to use, for ND routines that do not need
* multiple options of the same type.
*/
int
nd6_options(union nd_opts *ndopts)
{
struct nd_opt_hdr *nd_opt;
int i = 0;
KASSERT(ndopts != NULL, ("%s: ndopts == NULL", __func__));
KASSERT(ndopts->nd_opts_last != NULL, ("%s: uninitialized ndopts",
__func__));
if (ndopts->nd_opts_search == NULL)
return 0;
while (1) {
nd_opt = nd6_option(ndopts);
if (nd_opt == NULL && ndopts->nd_opts_last == NULL) {
/*
* Message validation requires that all included
* options have a length that is greater than zero.
*/
ICMP6STAT_INC(icp6s_nd_badopt);
bzero(ndopts, sizeof(*ndopts));
return -1;
}
if (nd_opt == NULL)
goto skip1;
switch (nd_opt->nd_opt_type) {
case ND_OPT_SOURCE_LINKADDR:
case ND_OPT_TARGET_LINKADDR:
case ND_OPT_MTU:
case ND_OPT_REDIRECTED_HEADER:
case ND_OPT_NONCE:
if (ndopts->nd_opt_array[nd_opt->nd_opt_type]) {
nd6log((LOG_INFO,
"duplicated ND6 option found (type=%d)\n",
nd_opt->nd_opt_type));
/* XXX bark? */
} else {
ndopts->nd_opt_array[nd_opt->nd_opt_type]
= nd_opt;
}
break;
case ND_OPT_PREFIX_INFORMATION:
if (ndopts->nd_opt_array[nd_opt->nd_opt_type] == 0) {
ndopts->nd_opt_array[nd_opt->nd_opt_type]
= nd_opt;
}
ndopts->nd_opts_pi_end =
(struct nd_opt_prefix_info *)nd_opt;
break;
/* What about ND_OPT_ROUTE_INFO? RFC 4191 */
case ND_OPT_RDNSS: /* RFC 6106 */
case ND_OPT_DNSSL: /* RFC 6106 */
/*
* Silently ignore options we know and do not care about
* in the kernel.
*/
break;
default:
/*
* Unknown options must be silently ignored,
* to accommodate future extension to the protocol.
*/
nd6log((LOG_DEBUG,
"nd6_options: unsupported option %d - "
"option ignored\n", nd_opt->nd_opt_type));
}
skip1:
i++;
if (i > V_nd6_maxndopt) {
ICMP6STAT_INC(icp6s_nd_toomanyopt);
nd6log((LOG_INFO, "too many loop in nd opt\n"));
break;
}
if (ndopts->nd_opts_done)
break;
}
return 0;
}
/*
* ND6 timer routine to handle ND6 entries
*/
static void
nd6_llinfo_settimer_locked(struct llentry *ln, long tick)
{
int canceled;
LLE_WLOCK_ASSERT(ln);
/* Do not schedule timers for child LLEs. */
if (ln->la_flags & LLE_CHILD)
return;
if (tick < 0) {
ln->la_expire = 0;
ln->ln_ntick = 0;
canceled = callout_stop(&ln->lle_timer);
} else {
ln->la_expire = time_uptime + tick / hz;
LLE_ADDREF(ln);
if (tick > INT_MAX) {
ln->ln_ntick = tick - INT_MAX;
canceled = callout_reset(&ln->lle_timer, INT_MAX,
nd6_llinfo_timer, ln);
} else {
ln->ln_ntick = 0;
canceled = callout_reset(&ln->lle_timer, tick,
nd6_llinfo_timer, ln);
}
}
if (canceled > 0)
LLE_REMREF(ln);
}
/*
* Gets source address of the first packet in hold queue
* and stores it in @src.
* Returns pointer to @src (if hold queue is not empty) or NULL.
*
* Set noinline to be dtrace-friendly
*/
static __noinline struct in6_addr *
nd6_llinfo_get_holdsrc(struct llentry *ln, struct in6_addr *src)
{
struct ip6_hdr hdr;
struct mbuf *m;
if (ln->la_hold == NULL)
return (NULL);
/*
* assume every packet in la_hold has the same IP header
*/
m = ln->la_hold;
if (sizeof(hdr) > m->m_len)
return (NULL);
m_copydata(m, 0, sizeof(hdr), (caddr_t)&hdr);
*src = hdr.ip6_src;
return (src);
}
/*
* Checks if we need to switch from STALE state.
*
* RFC 4861 requires switching from STALE to DELAY state
* on first packet matching entry, waiting V_nd6_delay and
* transition to PROBE state (if upper layer confirmation was
* not received).
*
* This code performs a bit differently:
* On packet hit we don't change state (but desired state
* can be guessed by control plane). However, after V_nd6_delay
* seconds code will transition to PROBE state (so DELAY state
* is kinda skipped in most situations).
*
* Typically, V_nd6_gctimer is bigger than V_nd6_delay, so
* we perform the following upon entering STALE state:
*
* 1) Arm timer to run each V_nd6_delay seconds to make sure that
* if packet was transmitted at the start of given interval, we
* would be able to switch to PROBE state in V_nd6_delay seconds
* as user expects.
*
* 2) Reschedule timer until original V_nd6_gctimer expires keeping
* lle in STALE state (remaining timer value stored in lle_remtime).
*
* 3) Reschedule timer if packet was transmitted less that V_nd6_delay
* seconds ago.
*
* Returns non-zero value if the entry is still STALE (storing
* the next timer interval in @pdelay).
*
* Returns zero value if original timer expired or we need to switch to
* PROBE (store that in @do_switch variable).
*/
static int
nd6_is_stale(struct llentry *lle, long *pdelay, int *do_switch)
{
int nd_delay, nd_gctimer;
time_t lle_hittime;
long delay;
*do_switch = 0;
nd_gctimer = V_nd6_gctimer;
nd_delay = V_nd6_delay;
lle_hittime = llentry_get_hittime(lle);
if (lle_hittime == 0) {
/*
* Datapath feedback has been requested upon entering
* STALE state. No packets has been passed using this lle.
* Ask for the timer reschedule and keep STALE state.
*/
delay = (long)(MIN(nd_gctimer, nd_delay));
delay *= hz;
if (lle->lle_remtime > delay)
lle->lle_remtime -= delay;
else {
delay = lle->lle_remtime;
lle->lle_remtime = 0;
}
if (delay == 0) {
/*
* The original ng6_gctime timeout ended,
* no more rescheduling.
*/
return (0);
}
*pdelay = delay;
return (1);
}
/*
* Packet received. Verify timestamp
*/
delay = (long)(time_uptime - lle_hittime);
if (delay < nd_delay) {
/*
* V_nd6_delay still not passed since the first
* hit in STALE state.
* Reschedule timer and return.
*/
*pdelay = (long)(nd_delay - delay) * hz;
return (1);
}
/* Request switching to probe */
*do_switch = 1;
return (0);
}
/*
* Switch @lle state to new state optionally arming timers.
*
* Set noinline to be dtrace-friendly
*/
__noinline void
nd6_llinfo_setstate(struct llentry *lle, int newstate)
{
struct ifnet *ifp;
int nd_gctimer, nd_delay;
long delay, remtime;
delay = 0;
remtime = 0;
switch (newstate) {
case ND6_LLINFO_INCOMPLETE:
ifp = lle->lle_tbl->llt_ifp;
delay = (long)ND_IFINFO(ifp)->retrans * hz / 1000;
break;
case ND6_LLINFO_REACHABLE:
if (!ND6_LLINFO_PERMANENT(lle)) {
ifp = lle->lle_tbl->llt_ifp;
delay = (long)ND_IFINFO(ifp)->reachable * hz;
}
break;
case ND6_LLINFO_STALE:
llentry_request_feedback(lle);
nd_delay = V_nd6_delay;
nd_gctimer = V_nd6_gctimer;
delay = (long)(MIN(nd_gctimer, nd_delay)) * hz;
remtime = (long)nd_gctimer * hz - delay;
break;
case ND6_LLINFO_DELAY:
lle->la_asked = 0;
delay = (long)V_nd6_delay * hz;
break;
}
if (delay > 0)
nd6_llinfo_settimer_locked(lle, delay);
lle->lle_remtime = remtime;
lle->ln_state = newstate;
}
/*
* Timer-dependent part of nd state machine.
*
* Set noinline to be dtrace-friendly
*/
static __noinline void
nd6_llinfo_timer(void *arg)
{
struct epoch_tracker et;
struct llentry *ln;
struct in6_addr *dst, *pdst, *psrc, src;
struct ifnet *ifp;
struct nd_ifinfo *ndi;
int do_switch, send_ns;
long delay;
KASSERT(arg != NULL, ("%s: arg NULL", __func__));
ln = (struct llentry *)arg;
ifp = lltable_get_ifp(ln->lle_tbl);
CURVNET_SET(ifp->if_vnet);
ND6_RLOCK();
LLE_WLOCK(ln);
if (callout_pending(&ln->lle_timer)) {
/*
* Here we are a bit odd here in the treatment of
* active/pending. If the pending bit is set, it got
* rescheduled before I ran. The active
* bit we ignore, since if it was stopped
* in ll_tablefree() and was currently running
* it would have return 0 so the code would
* not have deleted it since the callout could
* not be stopped so we want to go through
* with the delete here now. If the callout
* was restarted, the pending bit will be back on and
* we just want to bail since the callout_reset would
* return 1 and our reference would have been removed
* by nd6_llinfo_settimer_locked above since canceled
* would have been 1.
*/
LLE_WUNLOCK(ln);
ND6_RUNLOCK();
CURVNET_RESTORE();
return;
}
NET_EPOCH_ENTER(et);
ndi = ND_IFINFO(ifp);
send_ns = 0;
dst = &ln->r_l3addr.addr6;
pdst = dst;
if (ln->ln_ntick > 0) {
if (ln->ln_ntick > INT_MAX) {
ln->ln_ntick -= INT_MAX;
nd6_llinfo_settimer_locked(ln, INT_MAX);
} else {
ln->ln_ntick = 0;
nd6_llinfo_settimer_locked(ln, ln->ln_ntick);
}
goto done;
}
if (ln->la_flags & LLE_STATIC) {
goto done;
}
if (ln->la_flags & LLE_DELETED) {
nd6_free(&ln, 0);
goto done;
}
switch (ln->ln_state) {
case ND6_LLINFO_INCOMPLETE:
if (ln->la_asked < V_nd6_mmaxtries) {
ln->la_asked++;
send_ns = 1;
/* Send NS to multicast address */
pdst = NULL;
} else {
struct mbuf *m = ln->la_hold;
if (m) {
struct mbuf *m0;
/*
* assuming every packet in la_hold has the
* same IP header. Send error after unlock.
*/
m0 = m->m_nextpkt;
m->m_nextpkt = NULL;
ln->la_hold = m0;
clear_llinfo_pqueue(ln);
}
nd6_free(&ln, 0);
if (m != NULL) {
struct mbuf *n = m;
/*
* if there are any ummapped mbufs, we
* must free them, rather than using
* them for an ICMP, as they cannot be
* checksummed.
*/
while ((n = n->m_next) != NULL) {
if (n->m_flags & M_EXTPG)
break;
}
if (n != NULL) {
m_freem(m);
m = NULL;
} else {
icmp6_error2(m, ICMP6_DST_UNREACH,
ICMP6_DST_UNREACH_ADDR, 0, ifp);
}
}
}
break;
case ND6_LLINFO_REACHABLE:
if (!ND6_LLINFO_PERMANENT(ln))
nd6_llinfo_setstate(ln, ND6_LLINFO_STALE);
break;
case ND6_LLINFO_STALE:
if (nd6_is_stale(ln, &delay, &do_switch) != 0) {
/*
* No packet has used this entry and GC timeout
* has not been passed. Reschedule timer and
* return.
*/
nd6_llinfo_settimer_locked(ln, delay);
break;
}
if (do_switch == 0) {
/*
* GC timer has ended and entry hasn't been used.
* Run Garbage collector (RFC 4861, 5.3)
*/
if (!ND6_LLINFO_PERMANENT(ln))
nd6_free(&ln, 1);
break;
}
/* Entry has been used AND delay timer has ended. */
/* FALLTHROUGH */
case ND6_LLINFO_DELAY:
if (ndi && (ndi->flags & ND6_IFF_PERFORMNUD) != 0) {
/* We need NUD */
ln->la_asked = 1;
nd6_llinfo_setstate(ln, ND6_LLINFO_PROBE);
send_ns = 1;
} else
nd6_llinfo_setstate(ln, ND6_LLINFO_STALE); /* XXX */
break;
case ND6_LLINFO_PROBE:
if (ln->la_asked < V_nd6_umaxtries) {
ln->la_asked++;
send_ns = 1;
} else {
nd6_free(&ln, 0);
}
break;
default:
panic("%s: paths in a dark night can be confusing: %d",
__func__, ln->ln_state);
}
done:
if (ln != NULL)
ND6_RUNLOCK();
if (send_ns != 0) {
nd6_llinfo_settimer_locked(ln, (long)ndi->retrans * hz / 1000);
psrc = nd6_llinfo_get_holdsrc(ln, &src);
LLE_FREE_LOCKED(ln);
ln = NULL;
nd6_ns_output(ifp, psrc, pdst, dst, NULL);
}
if (ln != NULL)
LLE_FREE_LOCKED(ln);
NET_EPOCH_EXIT(et);
CURVNET_RESTORE();
}
/*
* ND6 timer routine to expire default route list and prefix list
*/
void
nd6_timer(void *arg)
{
CURVNET_SET((struct vnet *) arg);
struct epoch_tracker et;
struct nd_prhead prl;
struct nd_prefix *pr, *npr;
struct ifnet *ifp;
struct in6_ifaddr *ia6, *nia6;
uint64_t genid;
LIST_INIT(&prl);
NET_EPOCH_ENTER(et);
nd6_defrouter_timer();
/*
* expire interface addresses.
* in the past the loop was inside prefix expiry processing.
* However, from a stricter speci-confrmance standpoint, we should
* rather separate address lifetimes and prefix lifetimes.
*
* XXXRW: in6_ifaddrhead locking.
*/
addrloop:
CK_STAILQ_FOREACH_SAFE(ia6, &V_in6_ifaddrhead, ia_link, nia6) {
/* check address lifetime */
if (IFA6_IS_INVALID(ia6)) {
int regen = 0;
/*
* If the expiring address is temporary, try
* regenerating a new one. This would be useful when
* we suspended a laptop PC, then turned it on after a
* period that could invalidate all temporary
* addresses. Although we may have to restart the
* loop (see below), it must be after purging the
* address. Otherwise, we'd see an infinite loop of
* regeneration.
*/
if (V_ip6_use_tempaddr &&
(ia6->ia6_flags & IN6_IFF_TEMPORARY) != 0) {
if (regen_tmpaddr(ia6) == 0)
regen = 1;
}
in6_purgeaddr(&ia6->ia_ifa);
if (regen)
goto addrloop; /* XXX: see below */
} else if (IFA6_IS_DEPRECATED(ia6)) {
int oldflags = ia6->ia6_flags;
ia6->ia6_flags |= IN6_IFF_DEPRECATED;
/*
* If a temporary address has just become deprecated,
* regenerate a new one if possible.
*/
if (V_ip6_use_tempaddr &&
(ia6->ia6_flags & IN6_IFF_TEMPORARY) != 0 &&
(oldflags & IN6_IFF_DEPRECATED) == 0) {
if (regen_tmpaddr(ia6) == 0) {
/*
* A new temporary address is
* generated.
* XXX: this means the address chain
* has changed while we are still in
* the loop. Although the change
* would not cause disaster (because
* it's not a deletion, but an
* addition,) we'd rather restart the
* loop just for safety. Or does this
* significantly reduce performance??
*/
goto addrloop;
}
}
} else if ((ia6->ia6_flags & IN6_IFF_TENTATIVE) != 0) {
/*
* Schedule DAD for a tentative address. This happens
* if the interface was down or not running
* when the address was configured.
*/
int delay;
delay = arc4random() %
(MAX_RTR_SOLICITATION_DELAY * hz);
nd6_dad_start((struct ifaddr *)ia6, delay);
} else {
/*
* Check status of the interface. If it is down,
* mark the address as tentative for future DAD.
*/
ifp = ia6->ia_ifp;
if ((ND_IFINFO(ifp)->flags & ND6_IFF_NO_DAD) == 0 &&
((ifp->if_flags & IFF_UP) == 0 ||
(ifp->if_drv_flags & IFF_DRV_RUNNING) == 0 ||
(ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) != 0)){
ia6->ia6_flags &= ~IN6_IFF_DUPLICATED;
ia6->ia6_flags |= IN6_IFF_TENTATIVE;
}
/*
* A new RA might have made a deprecated address
* preferred.
*/
ia6->ia6_flags &= ~IN6_IFF_DEPRECATED;
}
}
NET_EPOCH_EXIT(et);
ND6_WLOCK();
restart:
LIST_FOREACH_SAFE(pr, &V_nd_prefix, ndpr_entry, npr) {
/*
* Expire prefixes. Since the pltime is only used for
* autoconfigured addresses, pltime processing for prefixes is
* not necessary.
*
* Only unlink after all derived addresses have expired. This
* may not occur until two hours after the prefix has expired
* per RFC 4862. If the prefix expires before its derived
* addresses, mark it off-link. This will be done automatically
* after unlinking if no address references remain.
*/
if (pr->ndpr_vltime == ND6_INFINITE_LIFETIME ||
time_uptime - pr->ndpr_lastupdate <= pr->ndpr_vltime)
continue;
if (pr->ndpr_addrcnt == 0) {
nd6_prefix_unlink(pr, &prl);
continue;
}
if ((pr->ndpr_stateflags & NDPRF_ONLINK) != 0) {
genid = V_nd6_list_genid;
nd6_prefix_ref(pr);
ND6_WUNLOCK();
ND6_ONLINK_LOCK();
(void)nd6_prefix_offlink(pr);
ND6_ONLINK_UNLOCK();
ND6_WLOCK();
nd6_prefix_rele(pr);
if (genid != V_nd6_list_genid)
goto restart;
}
}
ND6_WUNLOCK();
while ((pr = LIST_FIRST(&prl)) != NULL) {
LIST_REMOVE(pr, ndpr_entry);
nd6_prefix_del(pr);
}
callout_reset(&V_nd6_timer_ch, V_nd6_prune * hz,
nd6_timer, curvnet);
CURVNET_RESTORE();
}
/*
* ia6 - deprecated/invalidated temporary address
*/
static int
regen_tmpaddr(struct in6_ifaddr *ia6)
{
struct ifaddr *ifa;
struct ifnet *ifp;
struct in6_ifaddr *public_ifa6 = NULL;
NET_EPOCH_ASSERT();
ifp = ia6->ia_ifa.ifa_ifp;
CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
struct in6_ifaddr *it6;
if (ifa->ifa_addr->sa_family != AF_INET6)
continue;
it6 = (struct in6_ifaddr *)ifa;
/* ignore no autoconf addresses. */
if ((it6->ia6_flags & IN6_IFF_AUTOCONF) == 0)
continue;
/* ignore autoconf addresses with different prefixes. */
if (it6->ia6_ndpr == NULL || it6->ia6_ndpr != ia6->ia6_ndpr)
continue;
/*
* Now we are looking at an autoconf address with the same
* prefix as ours. If the address is temporary and is still
* preferred, do not create another one. It would be rare, but
* could happen, for example, when we resume a laptop PC after
* a long period.
*/
if ((it6->ia6_flags & IN6_IFF_TEMPORARY) != 0 &&
!IFA6_IS_DEPRECATED(it6)) {
public_ifa6 = NULL;
break;
}
/*
* This is a public autoconf address that has the same prefix
* as ours. If it is preferred, keep it. We can't break the
* loop here, because there may be a still-preferred temporary
* address with the prefix.
*/
if (!IFA6_IS_DEPRECATED(it6))
public_ifa6 = it6;
}
if (public_ifa6 != NULL)
ifa_ref(&public_ifa6->ia_ifa);
if (public_ifa6 != NULL) {
int e;
if ((e = in6_tmpifadd(public_ifa6, 0, 0)) != 0) {
ifa_free(&public_ifa6->ia_ifa);
log(LOG_NOTICE, "regen_tmpaddr: failed to create a new"
" tmp addr,errno=%d\n", e);
return (-1);
}
ifa_free(&public_ifa6->ia_ifa);
return (0);
}
return (-1);
}
/*
* Remove prefix and default router list entries corresponding to ifp. Neighbor
* cache entries are freed in in6_domifdetach().
*/
void
nd6_purge(struct ifnet *ifp)
{
struct nd_prhead prl;
struct nd_prefix *pr, *npr;
LIST_INIT(&prl);
/* Purge default router list entries toward ifp. */
nd6_defrouter_purge(ifp);
ND6_WLOCK();
/*
* Remove prefixes on ifp. We should have already removed addresses on
* this interface, so no addresses should be referencing these prefixes.
*/
LIST_FOREACH_SAFE(pr, &V_nd_prefix, ndpr_entry, npr) {
if (pr->ndpr_ifp == ifp)
nd6_prefix_unlink(pr, &prl);
}
ND6_WUNLOCK();
/* Delete the unlinked prefix objects. */
while ((pr = LIST_FIRST(&prl)) != NULL) {
LIST_REMOVE(pr, ndpr_entry);
nd6_prefix_del(pr);
}
/* cancel default outgoing interface setting */
if (V_nd6_defifindex == ifp->if_index)
nd6_setdefaultiface(0);
if (ND_IFINFO(ifp)->flags & ND6_IFF_ACCEPT_RTADV) {
/* Refresh default router list. */
defrouter_select_fib(ifp->if_fib);
}
}
/*
* the caller acquires and releases the lock on the lltbls
* Returns the llentry locked
*/
struct llentry *
nd6_lookup(const struct in6_addr *addr6, int flags, struct ifnet *ifp)
{
struct sockaddr_in6 sin6;
struct llentry *ln;
bzero(&sin6, sizeof(sin6));
sin6.sin6_len = sizeof(struct sockaddr_in6);
sin6.sin6_family = AF_INET6;
sin6.sin6_addr = *addr6;
IF_AFDATA_LOCK_ASSERT(ifp);
ln = lla_lookup(LLTABLE6(ifp), flags, (struct sockaddr *)&sin6);
return (ln);
}
static struct llentry *
nd6_alloc(const struct in6_addr *addr6, int flags, struct ifnet *ifp)
{
struct sockaddr_in6 sin6;
struct llentry *ln;
bzero(&sin6, sizeof(sin6));
sin6.sin6_len = sizeof(struct sockaddr_in6);
sin6.sin6_family = AF_INET6;
sin6.sin6_addr = *addr6;
ln = lltable_alloc_entry(LLTABLE6(ifp), 0, (struct sockaddr *)&sin6);
if (ln != NULL)
ln->ln_state = ND6_LLINFO_NOSTATE;
return (ln);
}
/*
* Test whether a given IPv6 address is a neighbor or not, ignoring
* the actual neighbor cache. The neighbor cache is ignored in order
* to not reenter the routing code from within itself.
*/
static int
nd6_is_new_addr_neighbor(const struct sockaddr_in6 *addr, struct ifnet *ifp)
{
struct nd_prefix *pr;
struct ifaddr *ifa;
struct rt_addrinfo info;
struct sockaddr_in6 rt_key;
const struct sockaddr *dst6;
uint64_t genid;
int error, fibnum;
/*
* A link-local address is always a neighbor.
* XXX: a link does not necessarily specify a single interface.
*/
if (IN6_IS_ADDR_LINKLOCAL(&addr->sin6_addr)) {
struct sockaddr_in6 sin6_copy;
u_int32_t zone;
/*
* We need sin6_copy since sa6_recoverscope() may modify the
* content (XXX).
*/
sin6_copy = *addr;
if (sa6_recoverscope(&sin6_copy))
return (0); /* XXX: should be impossible */
if (in6_setscope(&sin6_copy.sin6_addr, ifp, &zone))
return (0);
if (sin6_copy.sin6_scope_id == zone)
return (1);
else
return (0);
}
bzero(&rt_key, sizeof(rt_key));
bzero(&info, sizeof(info));
info.rti_info[RTAX_DST] = (struct sockaddr *)&rt_key;
/*
* If the address matches one of our addresses,
* it should be a neighbor.
* If the address matches one of our on-link prefixes, it should be a
* neighbor.
*/
ND6_RLOCK();
restart:
LIST_FOREACH(pr, &V_nd_prefix, ndpr_entry) {
if (pr->ndpr_ifp != ifp)
continue;
if ((pr->ndpr_stateflags & NDPRF_ONLINK) == 0) {
dst6 = (const struct sockaddr *)&pr->ndpr_prefix;
/*
* We only need to check all FIBs if add_addr_allfibs
* is unset. If set, checking any FIB will suffice.
*/
fibnum = V_rt_add_addr_allfibs ? rt_numfibs - 1 : 0;
for (; fibnum < rt_numfibs; fibnum++) {
genid = V_nd6_list_genid;
ND6_RUNLOCK();
/*
* Restore length field before
* retrying lookup
*/
rt_key.sin6_len = sizeof(rt_key);
error = rib_lookup_info(fibnum, dst6, 0, 0,
&info);
ND6_RLOCK();
if (genid != V_nd6_list_genid)
goto restart;
if (error == 0)
break;
}
if (error != 0)
continue;
/*
* This is the case where multiple interfaces
* have the same prefix, but only one is installed
* into the routing table and that prefix entry
* is not the one being examined here.
*/
if (!IN6_ARE_ADDR_EQUAL(&pr->ndpr_prefix.sin6_addr,
&rt_key.sin6_addr))
continue;
}
if (IN6_ARE_MASKED_ADDR_EQUAL(&pr->ndpr_prefix.sin6_addr,
&addr->sin6_addr, &pr->ndpr_mask)) {
ND6_RUNLOCK();
return (1);
}
}
ND6_RUNLOCK();
/*
* If the address is assigned on the node of the other side of
* a p2p interface, the address should be a neighbor.
*/
if (ifp->if_flags & IFF_POINTOPOINT) {
struct epoch_tracker et;
NET_EPOCH_ENTER(et);
CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
if (ifa->ifa_addr->sa_family != addr->sin6_family)
continue;
if (ifa->ifa_dstaddr != NULL &&
sa_equal(addr, ifa->ifa_dstaddr)) {
NET_EPOCH_EXIT(et);
return 1;
}
}
NET_EPOCH_EXIT(et);
}
/*
* If the default router list is empty, all addresses are regarded
* as on-link, and thus, as a neighbor.
*/
if (ND_IFINFO(ifp)->flags & ND6_IFF_ACCEPT_RTADV &&
nd6_defrouter_list_empty() &&
V_nd6_defifindex == ifp->if_index) {
return (1);
}
return (0);
}
/*
* Detect if a given IPv6 address identifies a neighbor on a given link.
* XXX: should take care of the destination of a p2p link?
*/
int
nd6_is_addr_neighbor(const struct sockaddr_in6 *addr, struct ifnet *ifp)
{
struct llentry *lle;
int rc = 0;
NET_EPOCH_ASSERT();
IF_AFDATA_UNLOCK_ASSERT(ifp);
if (nd6_is_new_addr_neighbor(addr, ifp))
return (1);
/*
* Even if the address matches none of our addresses, it might be
* in the neighbor cache.
*/
if ((lle = nd6_lookup(&addr->sin6_addr, LLE_SF(AF_INET6, 0), ifp)) != NULL) {
LLE_RUNLOCK(lle);
rc = 1;
}
return (rc);
}
static __noinline void
nd6_free_children(struct llentry *lle)
{
struct llentry *child_lle;
NET_EPOCH_ASSERT();
LLE_WLOCK_ASSERT(lle);
while ((child_lle = CK_SLIST_FIRST(&lle->lle_children)) != NULL) {
LLE_WLOCK(child_lle);
lltable_unlink_child_entry(child_lle);
llentry_free(child_lle);
}
}
/*
* Tries to update @lle address/prepend data with new @lladdr.
*
* Returns true on success.
* In any case, @lle is returned wlocked.
*/
static __noinline bool
nd6_try_set_entry_addr_locked(struct ifnet *ifp, struct llentry *lle, char *lladdr)
{
u_char buf[LLE_MAX_LINKHDR];
int fam, off;
size_t sz;
sz = sizeof(buf);
if (lltable_calc_llheader(ifp, AF_INET6, lladdr, buf, &sz, &off) != 0)
return (false);
/* Update data */
lltable_set_entry_addr(ifp, lle, buf, sz, off);
struct llentry *child_lle;
CK_SLIST_FOREACH(child_lle, &lle->lle_children, lle_child_next) {
LLE_WLOCK(child_lle);
fam = child_lle->r_family;
sz = sizeof(buf);
if (lltable_calc_llheader(ifp, fam, lladdr, buf, &sz, &off) == 0) {
/* success */
lltable_set_entry_addr(ifp, child_lle, buf, sz, off);
child_lle->ln_state = ND6_LLINFO_REACHABLE;
}
LLE_WUNLOCK(child_lle);
}
return (true);
}
bool
nd6_try_set_entry_addr(struct ifnet *ifp, struct llentry *lle, char *lladdr)
{
NET_EPOCH_ASSERT();
LLE_WLOCK_ASSERT(lle);
if (!lltable_acquire_wlock(ifp, lle))
return (false);
bool ret = nd6_try_set_entry_addr_locked(ifp, lle, lladdr);
IF_AFDATA_WUNLOCK(ifp);
return (ret);
}
/*
* Free an nd6 llinfo entry.
* Since the function would cause significant changes in the kernel, DO NOT
* make it global, unless you have a strong reason for the change, and are sure
* that the change is safe.
*
* Set noinline to be dtrace-friendly
*/
static __noinline void
nd6_free(struct llentry **lnp, int gc)
{
struct ifnet *ifp;
struct llentry *ln;
struct nd_defrouter *dr;
ln = *lnp;
*lnp = NULL;
LLE_WLOCK_ASSERT(ln);
ND6_RLOCK_ASSERT();
KASSERT((ln->la_flags & LLE_CHILD) == 0, ("child lle"));
ifp = lltable_get_ifp(ln->lle_tbl);
if ((ND_IFINFO(ifp)->flags & ND6_IFF_ACCEPT_RTADV) != 0)
dr = defrouter_lookup_locked(&ln->r_l3addr.addr6, ifp);
else
dr = NULL;
ND6_RUNLOCK();
if ((ln->la_flags & LLE_DELETED) == 0)
EVENTHANDLER_INVOKE(lle_event, ln, LLENTRY_EXPIRED);
/*
* we used to have pfctlinput(PRC_HOSTDEAD) here.
* even though it is not harmful, it was not really necessary.
*/
/* cancel timer */
nd6_llinfo_settimer_locked(ln, -1);
if (ND_IFINFO(ifp)->flags & ND6_IFF_ACCEPT_RTADV) {
if (dr != NULL && dr->expire &&
ln->ln_state == ND6_LLINFO_STALE && gc) {
/*
* If the reason for the deletion is just garbage
* collection, and the neighbor is an active default
* router, do not delete it. Instead, reset the GC
* timer using the router's lifetime.
* Simply deleting the entry would affect default
* router selection, which is not necessarily a good
* thing, especially when we're using router preference
* values.
* XXX: the check for ln_state would be redundant,
* but we intentionally keep it just in case.
*/
if (dr->expire > time_uptime)
nd6_llinfo_settimer_locked(ln,
(dr->expire - time_uptime) * hz);
else
nd6_llinfo_settimer_locked(ln,
(long)V_nd6_gctimer * hz);
LLE_REMREF(ln);
LLE_WUNLOCK(ln);
defrouter_rele(dr);
return;
}
if (dr) {
/*
* Unreachablity of a router might affect the default
* router selection and on-link detection of advertised
* prefixes.
*/
/*
* Temporarily fake the state to choose a new default
* router and to perform on-link determination of
* prefixes correctly.
* Below the state will be set correctly,
* or the entry itself will be deleted.
*/
ln->ln_state = ND6_LLINFO_INCOMPLETE;
}
if (ln->ln_router || dr) {
/*
* We need to unlock to avoid a LOR with rt6_flush() with the
* rnh and for the calls to pfxlist_onlink_check() and
* defrouter_select_fib() in the block further down for calls
* into nd6_lookup(). We still hold a ref.
*/
LLE_WUNLOCK(ln);
/*
* rt6_flush must be called whether or not the neighbor
* is in the Default Router List.
* See a corresponding comment in nd6_na_input().
*/
rt6_flush(&ln->r_l3addr.addr6, ifp);
}
if (dr) {
/*
* Since defrouter_select_fib() does not affect the
* on-link determination and MIP6 needs the check
* before the default router selection, we perform
* the check now.
*/
pfxlist_onlink_check();
/*
* Refresh default router list.
*/
defrouter_select_fib(dr->ifp->if_fib);
}
/*
* If this entry was added by an on-link redirect, remove the
* corresponding host route.
*/
if (ln->la_flags & LLE_REDIRECT)
nd6_free_redirect(ln);
if (ln->ln_router || dr)
LLE_WLOCK(ln);
}
/*
* Save to unlock. We still hold an extra reference and will not
* free(9) in llentry_free() if someone else holds one as well.
*/
LLE_WUNLOCK(ln);
IF_AFDATA_LOCK(ifp);
LLE_WLOCK(ln);
/* Guard against race with other llentry_free(). */
if (ln->la_flags & LLE_LINKED) {
/* Remove callout reference */
LLE_REMREF(ln);
lltable_unlink_entry(ln->lle_tbl, ln);
}
IF_AFDATA_UNLOCK(ifp);
nd6_free_children(ln);
llentry_free(ln);
if (dr != NULL)
defrouter_rele(dr);
}
static int
nd6_isdynrte(const struct rtentry *rt, const struct nhop_object *nh, void *xap)
{
if (nh->nh_flags & NHF_REDIRECT)
return (1);
return (0);
}
/*
* Remove the rtentry for the given llentry,
* both of which were installed by a redirect.
*/
static void
nd6_free_redirect(const struct llentry *ln)
{
int fibnum;
struct sockaddr_in6 sin6;
struct rt_addrinfo info;
struct rib_cmd_info rc;
struct epoch_tracker et;
lltable_fill_sa_entry(ln, (struct sockaddr *)&sin6);
memset(&info, 0, sizeof(info));
info.rti_info[RTAX_DST] = (struct sockaddr *)&sin6;
info.rti_filter = nd6_isdynrte;
NET_EPOCH_ENTER(et);
for (fibnum = 0; fibnum < rt_numfibs; fibnum++)
rib_action(fibnum, RTM_DELETE, &info, &rc);
NET_EPOCH_EXIT(et);
}
/*
* Updates status of the default router route.
*/
static void
check_release_defrouter(struct rib_cmd_info *rc, void *_cbdata)
{
struct nd_defrouter *dr;
struct nhop_object *nh;
nh = rc->rc_nh_old;
if ((nh != NULL) && (nh->nh_flags & NHF_DEFAULT)) {
dr = defrouter_lookup(&nh->gw6_sa.sin6_addr, nh->nh_ifp);
if (dr != NULL) {
dr->installed = 0;
defrouter_rele(dr);
}
}
}
void
nd6_subscription_cb(struct rib_head *rnh, struct rib_cmd_info *rc, void *arg)
{
#ifdef ROUTE_MPATH
rib_decompose_notification(rc, check_release_defrouter, NULL);
#else
check_release_defrouter(rc, NULL);
#endif
}
int
nd6_ioctl(u_long cmd, caddr_t data, struct ifnet *ifp)
{
struct in6_ndireq *ndi = (struct in6_ndireq *)data;
struct in6_nbrinfo *nbi = (struct in6_nbrinfo *)data;
struct in6_ndifreq *ndif = (struct in6_ndifreq *)data;
struct epoch_tracker et;
int error = 0;
if (ifp->if_afdata[AF_INET6] == NULL)
return (EPFNOSUPPORT);
switch (cmd) {
case OSIOCGIFINFO_IN6:
#define ND ndi->ndi
/* XXX: old ndp(8) assumes a positive value for linkmtu. */
bzero(&ND, sizeof(ND));
ND.linkmtu = IN6_LINKMTU(ifp);
ND.maxmtu = ND_IFINFO(ifp)->maxmtu;
ND.basereachable = ND_IFINFO(ifp)->basereachable;
ND.reachable = ND_IFINFO(ifp)->reachable;
ND.retrans = ND_IFINFO(ifp)->retrans;
ND.flags = ND_IFINFO(ifp)->flags;
ND.recalctm = ND_IFINFO(ifp)->recalctm;
ND.chlim = ND_IFINFO(ifp)->chlim;
break;
case SIOCGIFINFO_IN6:
ND = *ND_IFINFO(ifp);
break;
case SIOCSIFINFO_IN6:
/*
* used to change host variables from userland.
* intended for a use on router to reflect RA configurations.
*/
/* 0 means 'unspecified' */
if (ND.linkmtu != 0) {
if (ND.linkmtu < IPV6_MMTU ||
ND.linkmtu > IN6_LINKMTU(ifp)) {
error = EINVAL;
break;
}
ND_IFINFO(ifp)->linkmtu = ND.linkmtu;
}
if (ND.basereachable != 0) {
int obasereachable = ND_IFINFO(ifp)->basereachable;
ND_IFINFO(ifp)->basereachable = ND.basereachable;
if (ND.basereachable != obasereachable)
ND_IFINFO(ifp)->reachable =
ND_COMPUTE_RTIME(ND.basereachable);
}
if (ND.retrans != 0)
ND_IFINFO(ifp)->retrans = ND.retrans;
if (ND.chlim != 0)
ND_IFINFO(ifp)->chlim = ND.chlim;
/* FALLTHROUGH */
case SIOCSIFINFO_FLAGS:
{
struct ifaddr *ifa;
struct in6_ifaddr *ia;
if ((ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) &&
!(ND.flags & ND6_IFF_IFDISABLED)) {
/* ifdisabled 1->0 transision */
/*
* If the interface is marked as ND6_IFF_IFDISABLED and
* has an link-local address with IN6_IFF_DUPLICATED,
* do not clear ND6_IFF_IFDISABLED.
* See RFC 4862, Section 5.4.5.
*/
NET_EPOCH_ENTER(et);
CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
if (ifa->ifa_addr->sa_family != AF_INET6)
continue;
ia = (struct in6_ifaddr *)ifa;
if ((ia->ia6_flags & IN6_IFF_DUPLICATED) &&
IN6_IS_ADDR_LINKLOCAL(IA6_IN6(ia)))
break;
}
NET_EPOCH_EXIT(et);
if (ifa != NULL) {
/* LLA is duplicated. */
ND.flags |= ND6_IFF_IFDISABLED;
log(LOG_ERR, "Cannot enable an interface"
" with a link-local address marked"
" duplicate.\n");
} else {
ND_IFINFO(ifp)->flags &= ~ND6_IFF_IFDISABLED;
if (ifp->if_flags & IFF_UP)
in6_if_up(ifp);
}
} else if (!(ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED) &&
(ND.flags & ND6_IFF_IFDISABLED)) {
/* ifdisabled 0->1 transision */
/* Mark all IPv6 address as tentative. */
ND_IFINFO(ifp)->flags |= ND6_IFF_IFDISABLED;
if (V_ip6_dad_count > 0 &&
(ND_IFINFO(ifp)->flags & ND6_IFF_NO_DAD) == 0) {
NET_EPOCH_ENTER(et);
CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead,
ifa_link) {
if (ifa->ifa_addr->sa_family !=
AF_INET6)
continue;
ia = (struct in6_ifaddr *)ifa;
ia->ia6_flags |= IN6_IFF_TENTATIVE;
}
NET_EPOCH_EXIT(et);
}
}
if (ND.flags & ND6_IFF_AUTO_LINKLOCAL) {
if (!(ND_IFINFO(ifp)->flags & ND6_IFF_AUTO_LINKLOCAL)) {
/* auto_linklocal 0->1 transision */
/* If no link-local address on ifp, configure */
ND_IFINFO(ifp)->flags |= ND6_IFF_AUTO_LINKLOCAL;
in6_ifattach(ifp, NULL);
} else if (!(ND.flags & ND6_IFF_IFDISABLED) &&
ifp->if_flags & IFF_UP) {
/*
* When the IF already has
* ND6_IFF_AUTO_LINKLOCAL, no link-local
* address is assigned, and IFF_UP, try to
* assign one.
*/
NET_EPOCH_ENTER(et);
CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead,
ifa_link) {
if (ifa->ifa_addr->sa_family !=
AF_INET6)
continue;
ia = (struct in6_ifaddr *)ifa;
if (IN6_IS_ADDR_LINKLOCAL(IA6_IN6(ia)))
break;
}
NET_EPOCH_EXIT(et);
if (ifa != NULL)
/* No LLA is configured. */
in6_ifattach(ifp, NULL);
}
}
ND_IFINFO(ifp)->flags = ND.flags;
break;
}
#undef ND
case SIOCSNDFLUSH_IN6: /* XXX: the ioctl name is confusing... */
/* sync kernel routing table with the default router list */
defrouter_reset();
defrouter_select_fib(RT_ALL_FIBS);
break;
case SIOCSPFXFLUSH_IN6:
{
/* flush all the prefix advertised by routers */
struct in6_ifaddr *ia, *ia_next;
struct nd_prefix *pr, *next;
struct nd_prhead prl;
LIST_INIT(&prl);
ND6_WLOCK();
LIST_FOREACH_SAFE(pr, &V_nd_prefix, ndpr_entry, next) {
if (IN6_IS_ADDR_LINKLOCAL(&pr->ndpr_prefix.sin6_addr))
continue; /* XXX */
nd6_prefix_unlink(pr, &prl);
}
ND6_WUNLOCK();
while ((pr = LIST_FIRST(&prl)) != NULL) {
LIST_REMOVE(pr, ndpr_entry);
/* XXXRW: in6_ifaddrhead locking. */
CK_STAILQ_FOREACH_SAFE(ia, &V_in6_ifaddrhead, ia_link,
ia_next) {
if ((ia->ia6_flags & IN6_IFF_AUTOCONF) == 0)
continue;
if (ia->ia6_ndpr == pr)
in6_purgeaddr(&ia->ia_ifa);
}
nd6_prefix_del(pr);
}
break;
}
case SIOCSRTRFLUSH_IN6:
{
/* flush all the default routers */
defrouter_reset();
nd6_defrouter_flush_all();
defrouter_select_fib(RT_ALL_FIBS);
break;
}
case SIOCGNBRINFO_IN6:
{
struct llentry *ln;
struct in6_addr nb_addr = nbi->addr; /* make local for safety */
if ((error = in6_setscope(&nb_addr, ifp, NULL)) != 0)
return (error);
NET_EPOCH_ENTER(et);
ln = nd6_lookup(&nb_addr, LLE_SF(AF_INET6, 0), ifp);
NET_EPOCH_EXIT(et);
if (ln == NULL) {
error = EINVAL;
break;
}
nbi->state = ln->ln_state;
nbi->asked = ln->la_asked;
nbi->isrouter = ln->ln_router;
if (ln->la_expire == 0)
nbi->expire = 0;
else
nbi->expire = ln->la_expire + ln->lle_remtime / hz +
(time_second - time_uptime);
LLE_RUNLOCK(ln);
break;
}
case SIOCGDEFIFACE_IN6: /* XXX: should be implemented as a sysctl? */
ndif->ifindex = V_nd6_defifindex;
break;
case SIOCSDEFIFACE_IN6: /* XXX: should be implemented as a sysctl? */
return (nd6_setdefaultiface(ndif->ifindex));
}
return (error);
}
/*
* Calculates new isRouter value based on provided parameters and
* returns it.
*/
static int
nd6_is_router(int type, int code, int is_new, int old_addr, int new_addr,
int ln_router)
{
/*
* ICMP6 type dependent behavior.
*
* NS: clear IsRouter if new entry
* RS: clear IsRouter
* RA: set IsRouter if there's lladdr
* redir: clear IsRouter if new entry
*
* RA case, (1):
* The spec says that we must set IsRouter in the following cases:
* - If lladdr exist, set IsRouter. This means (1-5).
* - If it is old entry (!newentry), set IsRouter. This means (7).
* So, based on the spec, in (1-5) and (7) cases we must set IsRouter.
* A quetion arises for (1) case. (1) case has no lladdr in the
* neighbor cache, this is similar to (6).
* This case is rare but we figured that we MUST NOT set IsRouter.
*
* is_new old_addr new_addr NS RS RA redir
* D R
* 0 n n (1) c ? s
* 0 y n (2) c s s
* 0 n y (3) c s s
* 0 y y (4) c s s
* 0 y y (5) c s s
* 1 -- n (6) c c c s
* 1 -- y (7) c c s c s
*
* (c=clear s=set)
*/
switch (type & 0xff) {
case ND_NEIGHBOR_SOLICIT:
/*
* New entry must have is_router flag cleared.
*/
if (is_new) /* (6-7) */
ln_router = 0;
break;
case ND_REDIRECT:
/*
* If the icmp is a redirect to a better router, always set the
* is_router flag. Otherwise, if the entry is newly created,
* clear the flag. [RFC 2461, sec 8.3]
*/
if (code == ND_REDIRECT_ROUTER)
ln_router = 1;
else {
if (is_new) /* (6-7) */
ln_router = 0;
}
break;
case ND_ROUTER_SOLICIT:
/*
* is_router flag must always be cleared.
*/
ln_router = 0;
break;
case ND_ROUTER_ADVERT:
/*
* Mark an entry with lladdr as a router.
*/
if ((!is_new && (old_addr || new_addr)) || /* (2-5) */
(is_new && new_addr)) { /* (7) */
ln_router = 1;
}
break;
}
return (ln_router);
}
/*
* Create neighbor cache entry and cache link-layer address,
* on reception of inbound ND6 packets. (RS/RA/NS/redirect)
*
* type - ICMP6 type
* code - type dependent information
*
*/
void
nd6_cache_lladdr(struct ifnet *ifp, struct in6_addr *from, char *lladdr,
int lladdrlen, int type, int code)
{
struct llentry *ln = NULL, *ln_tmp;
int is_newentry;
int do_update;
int olladdr;
int llchange;
int flags;
uint16_t router = 0;
struct mbuf *chain = NULL;
u_char linkhdr[LLE_MAX_LINKHDR];
size_t linkhdrsize;
int lladdr_off;
NET_EPOCH_ASSERT();
IF_AFDATA_UNLOCK_ASSERT(ifp);
KASSERT(ifp != NULL, ("%s: ifp == NULL", __func__));
KASSERT(from != NULL, ("%s: from == NULL", __func__));
/* nothing must be updated for unspecified address */
if (IN6_IS_ADDR_UNSPECIFIED(from))
return;
/*
* Validation about ifp->if_addrlen and lladdrlen must be done in
* the caller.
*
* XXX If the link does not have link-layer adderss, what should
* we do? (ifp->if_addrlen == 0)
* Spec says nothing in sections for RA, RS and NA. There's small
* description on it in NS section (RFC 2461 7.2.3).
*/
flags = lladdr ? LLE_EXCLUSIVE : 0;
ln = nd6_lookup(from, LLE_SF(AF_INET6, flags), ifp);
is_newentry = 0;
if (ln == NULL) {
flags |= LLE_EXCLUSIVE;
ln = nd6_alloc(from, 0, ifp);
if (ln == NULL)
return;
/*
* Since we already know all the data for the new entry,
* fill it before insertion.
*/
if (lladdr != NULL) {
linkhdrsize = sizeof(linkhdr);
if (lltable_calc_llheader(ifp, AF_INET6, lladdr,
- linkhdr, &linkhdrsize, &lladdr_off) != 0)
+ linkhdr, &linkhdrsize, &lladdr_off) != 0) {
+ lltable_free_entry(LLTABLE6(ifp), ln);
return;
+ }
lltable_set_entry_addr(ifp, ln, linkhdr, linkhdrsize,
lladdr_off);
}
IF_AFDATA_WLOCK(ifp);
LLE_WLOCK(ln);
/* Prefer any existing lle over newly-created one */
ln_tmp = nd6_lookup(from, LLE_SF(AF_INET6, LLE_EXCLUSIVE), ifp);
if (ln_tmp == NULL)
lltable_link_entry(LLTABLE6(ifp), ln);
IF_AFDATA_WUNLOCK(ifp);
if (ln_tmp == NULL) {
/* No existing lle, mark as new entry (6,7) */
is_newentry = 1;
if (lladdr != NULL) { /* (7) */
nd6_llinfo_setstate(ln, ND6_LLINFO_STALE);
EVENTHANDLER_INVOKE(lle_event, ln,
LLENTRY_RESOLVED);
}
} else {
lltable_free_entry(LLTABLE6(ifp), ln);
ln = ln_tmp;
ln_tmp = NULL;
}
}
/* do nothing if static ndp is set */
if ((ln->la_flags & LLE_STATIC)) {
if (flags & LLE_EXCLUSIVE)
LLE_WUNLOCK(ln);
else
LLE_RUNLOCK(ln);
return;
}
olladdr = (ln->la_flags & LLE_VALID) ? 1 : 0;
if (olladdr && lladdr) {
llchange = bcmp(lladdr, ln->ll_addr,
ifp->if_addrlen);
} else if (!olladdr && lladdr)
llchange = 1;
else
llchange = 0;
/*
* newentry olladdr lladdr llchange (*=record)
* 0 n n -- (1)
* 0 y n -- (2)
* 0 n y y (3) * STALE
* 0 y y n (4) *
* 0 y y y (5) * STALE
* 1 -- n -- (6) NOSTATE(= PASSIVE)
* 1 -- y -- (7) * STALE
*/
do_update = 0;
if (is_newentry == 0 && llchange != 0) {
do_update = 1; /* (3,5) */
/*
* Record source link-layer address
* XXX is it dependent to ifp->if_type?
*/
if (!nd6_try_set_entry_addr(ifp, ln, lladdr)) {
/* Entry was deleted */
LLE_WUNLOCK(ln);
return;
}
nd6_llinfo_setstate(ln, ND6_LLINFO_STALE);
EVENTHANDLER_INVOKE(lle_event, ln, LLENTRY_RESOLVED);
if (ln->la_hold != NULL)
chain = nd6_grab_holdchain(ln);
}
/* Calculates new router status */
router = nd6_is_router(type, code, is_newentry, olladdr,
lladdr != NULL ? 1 : 0, ln->ln_router);
ln->ln_router = router;
/* Mark non-router redirects with special flag */
if ((type & 0xFF) == ND_REDIRECT && code != ND_REDIRECT_ROUTER)
ln->la_flags |= LLE_REDIRECT;
if (flags & LLE_EXCLUSIVE)
LLE_WUNLOCK(ln);
else
LLE_RUNLOCK(ln);
if (chain != NULL)
nd6_flush_holdchain(ifp, ln, chain);
if (do_update)
nd6_flush_children_holdchain(ifp, ln);
/*
* When the link-layer address of a router changes, select the
* best router again. In particular, when the neighbor entry is newly
* created, it might affect the selection policy.
* Question: can we restrict the first condition to the "is_newentry"
* case?
* XXX: when we hear an RA from a new router with the link-layer
* address option, defrouter_select_fib() is called twice, since
* defrtrlist_update called the function as well. However, I believe
* we can compromise the overhead, since it only happens the first
* time.
* XXX: although defrouter_select_fib() should not have a bad effect
* for those are not autoconfigured hosts, we explicitly avoid such
* cases for safety.
*/
if ((do_update || is_newentry) && router &&
ND_IFINFO(ifp)->flags & ND6_IFF_ACCEPT_RTADV) {
/*
* guaranteed recursion
*/
defrouter_select_fib(ifp->if_fib);
}
}
static void
nd6_slowtimo(void *arg)
{
struct epoch_tracker et;
CURVNET_SET((struct vnet *) arg);
struct nd_ifinfo *nd6if;
struct ifnet *ifp;
callout_reset(&V_nd6_slowtimo_ch, ND6_SLOWTIMER_INTERVAL * hz,
nd6_slowtimo, curvnet);
NET_EPOCH_ENTER(et);
CK_STAILQ_FOREACH(ifp, &V_ifnet, if_link) {
if (ifp->if_afdata[AF_INET6] == NULL)
continue;
nd6if = ND_IFINFO(ifp);
if (nd6if->basereachable && /* already initialized */
(nd6if->recalctm -= ND6_SLOWTIMER_INTERVAL) <= 0) {
/*
* Since reachable time rarely changes by router
* advertisements, we SHOULD insure that a new random
* value gets recomputed at least once every few hours.
* (RFC 2461, 6.3.4)
*/
nd6if->recalctm = V_nd6_recalc_reachtm_interval;
nd6if->reachable = ND_COMPUTE_RTIME(nd6if->basereachable);
}
}
NET_EPOCH_EXIT(et);
CURVNET_RESTORE();
}
struct mbuf *
nd6_grab_holdchain(struct llentry *ln)
{
struct mbuf *chain;
LLE_WLOCK_ASSERT(ln);
chain = ln->la_hold;
ln->la_hold = NULL;
if (ln->ln_state == ND6_LLINFO_STALE) {
/*
* The first time we send a packet to a
* neighbor whose entry is STALE, we have
* to change the state to DELAY and a sets
* a timer to expire in DELAY_FIRST_PROBE_TIME
* seconds to ensure do neighbor unreachability
* detection on expiration.
* (RFC 2461 7.3.3)
*/
nd6_llinfo_setstate(ln, ND6_LLINFO_DELAY);
}
return (chain);
}
int
nd6_output_ifp(struct ifnet *ifp, struct ifnet *origifp, struct mbuf *m,
struct sockaddr_in6 *dst, struct route *ro)
{
int error;
int ip6len;
struct ip6_hdr *ip6;
struct m_tag *mtag;
#ifdef MAC
mac_netinet6_nd6_send(ifp, m);
#endif
/*
* If called from nd6_ns_output() (NS), nd6_na_output() (NA),
* icmp6_redirect_output() (REDIRECT) or from rip6_output() (RS, RA
* as handled by rtsol and rtadvd), mbufs will be tagged for SeND
* to be diverted to user space. When re-injected into the kernel,
* send_output() will directly dispatch them to the outgoing interface.
*/
if (send_sendso_input_hook != NULL) {
mtag = m_tag_find(m, PACKET_TAG_ND_OUTGOING, NULL);
if (mtag != NULL) {
ip6 = mtod(m, struct ip6_hdr *);
ip6len = sizeof(struct ip6_hdr) + ntohs(ip6->ip6_plen);
/* Use the SEND socket */
error = send_sendso_input_hook(m, ifp, SND_OUT,
ip6len);
/* -1 == no app on SEND socket */
if (error == 0 || error != -1)
return (error);
}
}
m_clrprotoflags(m); /* Avoid confusing lower layers. */
IP_PROBE(send, NULL, NULL, mtod(m, struct ip6_hdr *), ifp, NULL,
mtod(m, struct ip6_hdr *));
if ((ifp->if_flags & IFF_LOOPBACK) == 0)
origifp = ifp;
error = (*ifp->if_output)(origifp, m, (struct sockaddr *)dst, ro);
return (error);
}
/*
* Lookup link headerfor @sa_dst address. Stores found
* data in @desten buffer. Copy of lle ln_flags can be also
* saved in @pflags if @pflags is non-NULL.
*
* If destination LLE does not exists or lle state modification
* is required, call "slow" version.
*
* Return values:
* - 0 on success (address copied to buffer).
* - EWOULDBLOCK (no local error, but address is still unresolved)
* - other errors (alloc failure, etc)
*/
int
nd6_resolve(struct ifnet *ifp, int gw_flags, struct mbuf *m,
const struct sockaddr *sa_dst, u_char *desten, uint32_t *pflags,
struct llentry **plle)
{
struct llentry *ln = NULL;
const struct sockaddr_in6 *dst6;
NET_EPOCH_ASSERT();
if (pflags != NULL)
*pflags = 0;
dst6 = (const struct sockaddr_in6 *)sa_dst;
/* discard the packet if IPv6 operation is disabled on the interface */
if ((ND_IFINFO(ifp)->flags & ND6_IFF_IFDISABLED)) {
m_freem(m);
return (ENETDOWN); /* better error? */
}
if (m != NULL && m->m_flags & M_MCAST) {
switch (ifp->if_type) {
case IFT_ETHER:
case IFT_L2VLAN:
case IFT_BRIDGE:
ETHER_MAP_IPV6_MULTICAST(&dst6->sin6_addr,
desten);
return (0);
default:
m_freem(m);
return (EAFNOSUPPORT);
}
}
int family = gw_flags >> 16;
int lookup_flags = plle ? LLE_EXCLUSIVE : LLE_UNLOCKED;
ln = nd6_lookup(&dst6->sin6_addr, LLE_SF(family, lookup_flags), ifp);
if (ln != NULL && (ln->r_flags & RLLE_VALID) != 0) {
/* Entry found, let's copy lle info */
bcopy(ln->r_linkdata, desten, ln->r_hdrlen);
if (pflags != NULL)
*pflags = LLE_VALID | (ln->r_flags & RLLE_IFADDR);
llentry_provide_feedback(ln);
if (plle) {
LLE_ADDREF(ln);
*plle = ln;
LLE_WUNLOCK(ln);
}
return (0);
} else if (plle && ln)
LLE_WUNLOCK(ln);
return (nd6_resolve_slow(ifp, family, 0, m, dst6, desten, pflags, plle));
}
/*
* Finds or creates a new llentry for @addr and @family.
* Returns wlocked llentry or NULL.
*
*
* Child LLEs.
*
* Do not have their own state machine (gets marked as static)
* settimer bails out for child LLEs just in case.
*
* Locking order: parent lle gets locked first, chen goes the child.
*/
static __noinline struct llentry *
nd6_get_llentry(struct ifnet *ifp, const struct in6_addr *addr, int family)
{
struct llentry *child_lle = NULL;
struct llentry *lle, *lle_tmp;
lle = nd6_alloc(addr, 0, ifp);
if (lle != NULL && family != AF_INET6) {
child_lle = nd6_alloc(addr, 0, ifp);
if (child_lle == NULL) {
lltable_free_entry(LLTABLE6(ifp), lle);
return (NULL);
}
child_lle->r_family = family;
child_lle->la_flags |= LLE_CHILD | LLE_STATIC;
child_lle->ln_state = ND6_LLINFO_INCOMPLETE;
}
if (lle == NULL) {
char ip6buf[INET6_ADDRSTRLEN];
log(LOG_DEBUG,
"nd6_get_llentry: can't allocate llinfo for %s "
"(ln=%p)\n",
ip6_sprintf(ip6buf, addr), lle);
return (NULL);
}
IF_AFDATA_WLOCK(ifp);
LLE_WLOCK(lle);
/* Prefer any existing entry over newly-created one */
lle_tmp = nd6_lookup(addr, LLE_SF(AF_INET6, LLE_EXCLUSIVE), ifp);
if (lle_tmp == NULL)
lltable_link_entry(LLTABLE6(ifp), lle);
else {
lltable_free_entry(LLTABLE6(ifp), lle);
lle = lle_tmp;
}
if (child_lle != NULL) {
/* Check if child lle for the same family exists */
lle_tmp = llentry_lookup_family(lle, child_lle->r_family);
LLE_WLOCK(child_lle);
if (lle_tmp == NULL) {
/* Attach */
lltable_link_child_entry(lle, child_lle);
} else {
/* child lle already exists, free newly-created one */
lltable_free_entry(LLTABLE6(ifp), child_lle);
child_lle = lle_tmp;
}
LLE_WUNLOCK(lle);
lle = child_lle;
}
IF_AFDATA_WUNLOCK(ifp);
return (lle);
}
/*
* Do L2 address resolution for @sa_dst address. Stores found
* address in @desten buffer. Copy of lle ln_flags can be also
* saved in @pflags if @pflags is non-NULL.
*
* Heavy version.
* Function assume that destination LLE does not exist,
* is invalid or stale, so LLE_EXCLUSIVE lock needs to be acquired.
*
* Set noinline to be dtrace-friendly
*/
static __noinline int
nd6_resolve_slow(struct ifnet *ifp, int family, int flags, struct mbuf *m,
const struct sockaddr_in6 *dst, u_char *desten, uint32_t *pflags,
struct llentry **plle)
{
struct llentry *lle = NULL;
struct in6_addr *psrc, src;
int send_ns, ll_len;
char *lladdr;
NET_EPOCH_ASSERT();
/*
* Address resolution or Neighbor Unreachability Detection
* for the next hop.
* At this point, the destination of the packet must be a unicast
* or an anycast address(i.e. not a multicast).
*/
lle = nd6_lookup(&dst->sin6_addr, LLE_SF(family, LLE_EXCLUSIVE), ifp);
if ((lle == NULL) && nd6_is_addr_neighbor(dst, ifp)) {
/*
* Since nd6_is_addr_neighbor() internally calls nd6_lookup(),
* the condition below is not very efficient. But we believe
* it is tolerable, because this should be a rare case.
*/
lle = nd6_get_llentry(ifp, &dst->sin6_addr, family);
}
if (lle == NULL) {
m_freem(m);
return (ENOBUFS);
}
LLE_WLOCK_ASSERT(lle);
/*
* The first time we send a packet to a neighbor whose entry is
* STALE, we have to change the state to DELAY and a sets a timer to
* expire in DELAY_FIRST_PROBE_TIME seconds to ensure do
* neighbor unreachability detection on expiration.
* (RFC 2461 7.3.3)
*/
if ((!(lle->la_flags & LLE_CHILD)) && (lle->ln_state == ND6_LLINFO_STALE))
nd6_llinfo_setstate(lle, ND6_LLINFO_DELAY);
/*
* If the neighbor cache entry has a state other than INCOMPLETE
* (i.e. its link-layer address is already resolved), just
* send the packet.
*/
if (lle->ln_state > ND6_LLINFO_INCOMPLETE) {
if (flags & LLE_ADDRONLY) {
lladdr = lle->ll_addr;
ll_len = ifp->if_addrlen;
} else {
lladdr = lle->r_linkdata;
ll_len = lle->r_hdrlen;
}
bcopy(lladdr, desten, ll_len);
if (pflags != NULL)
*pflags = lle->la_flags;
if (plle) {
LLE_ADDREF(lle);
*plle = lle;
}
LLE_WUNLOCK(lle);
return (0);
}
/*
* There is a neighbor cache entry, but no ethernet address
* response yet. Append this latest packet to the end of the
* packet queue in the mbuf. When it exceeds nd6_maxqueuelen,
* the oldest packet in the queue will be removed.
*/
if (lle->la_hold != NULL) {
struct mbuf *m_hold;
int i;
i = 0;
for (m_hold = lle->la_hold; m_hold; m_hold = m_hold->m_nextpkt){
i++;
if (m_hold->m_nextpkt == NULL) {
m_hold->m_nextpkt = m;
break;
}
}
while (i >= V_nd6_maxqueuelen) {
m_hold = lle->la_hold;
lle->la_hold = lle->la_hold->m_nextpkt;
m_freem(m_hold);
i--;
}
} else {
lle->la_hold = m;
}
/*
* If there has been no NS for the neighbor after entering the
* INCOMPLETE state, send the first solicitation.
* Note that for newly-created lle la_asked will be 0,
* so we will transition from ND6_LLINFO_NOSTATE to
* ND6_LLINFO_INCOMPLETE state here.
*/
psrc = NULL;
send_ns = 0;
/* If we have child lle, switch to the parent to send NS */
if (lle->la_flags & LLE_CHILD) {
struct llentry *lle_parent = lle->lle_parent;
LLE_WUNLOCK(lle);
lle = lle_parent;
LLE_WLOCK(lle);
}
if (lle->la_asked == 0) {
lle->la_asked++;
send_ns = 1;
psrc = nd6_llinfo_get_holdsrc(lle, &src);
nd6_llinfo_setstate(lle, ND6_LLINFO_INCOMPLETE);
}
LLE_WUNLOCK(lle);
if (send_ns != 0)
nd6_ns_output(ifp, psrc, NULL, &dst->sin6_addr, NULL);
return (EWOULDBLOCK);
}
/*
* Do L2 address resolution for @sa_dst address. Stores found
* address in @desten buffer. Copy of lle ln_flags can be also
* saved in @pflags if @pflags is non-NULL.
*
* Return values:
* - 0 on success (address copied to buffer).
* - EWOULDBLOCK (no local error, but address is still unresolved)
* - other errors (alloc failure, etc)
*/
int
nd6_resolve_addr(struct ifnet *ifp, int flags, const struct sockaddr *dst,
char *desten, uint32_t *pflags)
{
int error;
flags |= LLE_ADDRONLY;
error = nd6_resolve_slow(ifp, AF_INET6, flags, NULL,
(const struct sockaddr_in6 *)dst, desten, pflags, NULL);
return (error);
}
int
nd6_flush_holdchain(struct ifnet *ifp, struct llentry *lle, struct mbuf *chain)
{
struct mbuf *m, *m_head;
struct sockaddr_in6 dst6;
int error = 0;
NET_EPOCH_ASSERT();
struct route_in6 ro = {
.ro_prepend = lle->r_linkdata,
.ro_plen = lle->r_hdrlen,
};
lltable_fill_sa_entry(lle, (struct sockaddr *)&dst6);
m_head = chain;
while (m_head) {
m = m_head;
m_head = m_head->m_nextpkt;
m->m_nextpkt = NULL;
error = nd6_output_ifp(ifp, ifp, m, &dst6, (struct route *)&ro);
}
/*
* XXX
* note that intermediate errors are blindly ignored
*/
return (error);
}
__noinline void
nd6_flush_children_holdchain(struct ifnet *ifp, struct llentry *lle)
{
struct llentry *child_lle;
struct mbuf *chain;
NET_EPOCH_ASSERT();
CK_SLIST_FOREACH(child_lle, &lle->lle_children, lle_child_next) {
LLE_WLOCK(child_lle);
chain = nd6_grab_holdchain(child_lle);
LLE_WUNLOCK(child_lle);
nd6_flush_holdchain(ifp, child_lle, chain);
}
}
static int
nd6_need_cache(struct ifnet *ifp)
{
/*
* XXX: we currently do not make neighbor cache on any interface
* other than Ethernet and GIF.
*
* RFC2893 says:
* - unidirectional tunnels needs no ND
*/
switch (ifp->if_type) {
case IFT_ETHER:
case IFT_IEEE1394:
case IFT_L2VLAN:
case IFT_INFINIBAND:
case IFT_BRIDGE:
case IFT_PROPVIRTUAL:
return (1);
default:
return (0);
}
}
/*
* Add pernament ND6 link-layer record for given
* interface address.
*
* Very similar to IPv4 arp_ifinit(), but:
* 1) IPv6 DAD is performed in different place
* 2) It is called by IPv6 protocol stack in contrast to
* arp_ifinit() which is typically called in SIOCSIFADDR
* driver ioctl handler.
*
*/
int
nd6_add_ifa_lle(struct in6_ifaddr *ia)
{
struct ifnet *ifp;
struct llentry *ln, *ln_tmp;
struct sockaddr *dst;
ifp = ia->ia_ifa.ifa_ifp;
if (nd6_need_cache(ifp) == 0)
return (0);
dst = (struct sockaddr *)&ia->ia_addr;
ln = lltable_alloc_entry(LLTABLE6(ifp), LLE_IFADDR, dst);
if (ln == NULL)
return (ENOBUFS);
IF_AFDATA_WLOCK(ifp);
LLE_WLOCK(ln);
/* Unlink any entry if exists */
ln_tmp = lla_lookup(LLTABLE6(ifp), LLE_SF(AF_INET6, LLE_EXCLUSIVE), dst);
if (ln_tmp != NULL)
lltable_unlink_entry(LLTABLE6(ifp), ln_tmp);
lltable_link_entry(LLTABLE6(ifp), ln);
IF_AFDATA_WUNLOCK(ifp);
if (ln_tmp != NULL)
EVENTHANDLER_INVOKE(lle_event, ln_tmp, LLENTRY_EXPIRED);
EVENTHANDLER_INVOKE(lle_event, ln, LLENTRY_RESOLVED);
LLE_WUNLOCK(ln);
if (ln_tmp != NULL)
llentry_free(ln_tmp);
return (0);
}
/*
* Removes either all lle entries for given @ia, or lle
* corresponding to @ia address.
*/
void
nd6_rem_ifa_lle(struct in6_ifaddr *ia, int all)
{
struct sockaddr_in6 mask, addr;
struct sockaddr *saddr, *smask;
struct ifnet *ifp;
ifp = ia->ia_ifa.ifa_ifp;
memcpy(&addr, &ia->ia_addr, sizeof(ia->ia_addr));
memcpy(&mask, &ia->ia_prefixmask, sizeof(ia->ia_prefixmask));
saddr = (struct sockaddr *)&addr;
smask = (struct sockaddr *)&mask;
if (all != 0)
lltable_prefix_free(AF_INET6, saddr, smask, LLE_STATIC);
else
lltable_delete_addr(LLTABLE6(ifp), LLE_IFADDR, saddr);
}
static void
clear_llinfo_pqueue(struct llentry *ln)
{
struct mbuf *m_hold, *m_hold_next;
for (m_hold = ln->la_hold; m_hold; m_hold = m_hold_next) {
m_hold_next = m_hold->m_nextpkt;
m_freem(m_hold);
}
ln->la_hold = NULL;
}
static int
nd6_sysctl_prlist(SYSCTL_HANDLER_ARGS)
{
struct in6_prefix p;
struct sockaddr_in6 s6;
struct nd_prefix *pr;
struct nd_pfxrouter *pfr;
time_t maxexpire;
int error;
char ip6buf[INET6_ADDRSTRLEN];
if (req->newptr)
return (EPERM);
error = sysctl_wire_old_buffer(req, 0);
if (error != 0)
return (error);
bzero(&p, sizeof(p));
p.origin = PR_ORIG_RA;
bzero(&s6, sizeof(s6));
s6.sin6_family = AF_INET6;
s6.sin6_len = sizeof(s6);
ND6_RLOCK();
LIST_FOREACH(pr, &V_nd_prefix, ndpr_entry) {
p.prefix = pr->ndpr_prefix;
if (sa6_recoverscope(&p.prefix)) {
log(LOG_ERR, "scope error in prefix list (%s)\n",
ip6_sprintf(ip6buf, &p.prefix.sin6_addr));
/* XXX: press on... */
}
p.raflags = pr->ndpr_raf;
p.prefixlen = pr->ndpr_plen;
p.vltime = pr->ndpr_vltime;
p.pltime = pr->ndpr_pltime;
p.if_index = pr->ndpr_ifp->if_index;
if (pr->ndpr_vltime == ND6_INFINITE_LIFETIME)
p.expire = 0;
else {
/* XXX: we assume time_t is signed. */
maxexpire = (-1) &
~((time_t)1 << ((sizeof(maxexpire) * 8) - 1));
if (pr->ndpr_vltime < maxexpire - pr->ndpr_lastupdate)
p.expire = pr->ndpr_lastupdate +
pr->ndpr_vltime +
(time_second - time_uptime);
else
p.expire = maxexpire;
}
p.refcnt = pr->ndpr_addrcnt;
p.flags = pr->ndpr_stateflags;
p.advrtrs = 0;
LIST_FOREACH(pfr, &pr->ndpr_advrtrs, pfr_entry)
p.advrtrs++;
error = SYSCTL_OUT(req, &p, sizeof(p));
if (error != 0)
break;
LIST_FOREACH(pfr, &pr->ndpr_advrtrs, pfr_entry) {
s6.sin6_addr = pfr->router->rtaddr;
if (sa6_recoverscope(&s6))
log(LOG_ERR,
"scope error in prefix list (%s)\n",
ip6_sprintf(ip6buf, &pfr->router->rtaddr));
error = SYSCTL_OUT(req, &s6, sizeof(s6));
if (error != 0)
goto out;
}
}
out:
ND6_RUNLOCK();
return (error);
}
SYSCTL_PROC(_net_inet6_icmp6, ICMPV6CTL_ND6_PRLIST, nd6_prlist,
CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_MPSAFE,
NULL, 0, nd6_sysctl_prlist, "S,in6_prefix",
"NDP prefix list");
SYSCTL_INT(_net_inet6_icmp6, ICMPV6CTL_ND6_MAXQLEN, nd6_maxqueuelen,
CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(nd6_maxqueuelen), 1, "");
SYSCTL_INT(_net_inet6_icmp6, OID_AUTO, nd6_gctimer,
CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(nd6_gctimer), (60 * 60 * 24), "");